Benthic Primary Production and Carbon Turnover in Coastal Marine Environments Quantified Using Aquatic Eddy Correlation

BENTHIC PRIMARY PRODUCTION AND CARBON TURNOVER IN COASTAL MARINE ENVIRONMENTS QUANTIFIED USING AQUATIC EDDY CORRELATION

Karl M. Attard

Ph.D. thesis

September 2014

1 Nordic Centre for Earth Evolution

BENTHIC PRIMARY PRODUCTION AND CARBON TURNOVER IN COASTAL MARINE ENVIRONMENTS QUANTIFIED USING AQUATIC EDDY CORRELATION

Karl M. Attard

A dissertation submitted to the Faculty of Science

at the University of Southern Denmark in partial

fulfilment of the requirements for the degree of

Doctor of Philosophy

Department of Biology

Nordic Centre for Earth Evolution

September 2014

3 Nordic Centre for Earth Evolution

Data sheet

Title: Benthic primary production and carbon turnover in coastal marine environments quantified using aquatic eddy correlation

Author: Karl M. Attard

Affiliation: Institute of Biology, Nordic Centre for Earth Evolution, University of Southern Denmark, and Greenland Climate Research Centre, Greenland Institute of Natural Resources

Supervisors: Prof. Ronnie N. Glud Asst. Prof. Daniel F. McGinnis Prof. Søren Rysgaard

Committee: Assoc. Prof. Erik Kristensen (University of Southern Denmark) Res. Prof. Peter Berg (University of Virginia) Prof. Per Hall (University of Gothenburg)

Submitted: September 2014

Cite as: Attard, K. M. 2014. Benthic primary production and carbon turnover in coastal marine environments quantified using aquatic eddy correlation. PhD thesis. University of Southern Denmark and Greenland Climate Research Centre. 197pp.

Keywords: Benthic oxygen exchange, benthic primary production, coastal carbon cycling, benthic metabolism, eddy correlation, Greenland, maerl beds, cold water corals, permeable sediments

Front page: (Left) Deploying an eddy correlation system by ROV on a cold water coral reef in the NE Atlantic during the JC073 ‘Changing Oceans’ expedition. (Right) The remarkable pristine maerl beds in Loch Sween, Scotland (Photo by Rob Cook, Heriot-Watt University).

Contents

Acknowledgements 09

Abstract 11

Dansk Resumé 13

List of manuscripts 15

1. INTRODUCTION

1.1. The oceanic carbon cycle 17

Benthic primary production 18

Light availability 19

Organic carbon degradation 20

Importance of benthic fauna 21

1.2. The benthic O2 exchange rate 24

The benthic O2 exchange rate as a measure of organic carbon remineralization 24

Estimating benthic primary production from O2 exchange measurements 26

Evaluating the benthic community light response 27

1.3. Estimating the benthic O2 exchange rate 30

The benthic boundary layer 30

Benthic O2 microprofile and chamber measurements 31

The aquatic O2 eddy correlation method 34

1.4. Summary of manuscripts 43

1.5. References 49

2. APPENDIX 57

3. PUBLICATIONS

3.1. Paper I 63

3.2. Paper II 113

3.3. Paper III 143

ACKNOWLEDGEMENTS

I must begin by acknowledging my main advisor, Prof. Ronnie N. Glud, who back in September 2010 saw it fit to offer me a Ph.D. position within his research group. Looking back at the past four years, I couldn’t have hoped for a better set of experiences, both academic and personal, than this Ph.D. project has provided me with. I thank him for being an excellent mentor and for always going the extra mile, whether it was his flexibility in scheduling or in providing honest advice and guidance. I thank my co-adviser, Dr. Daniel F. McGinnis, for introducing me to the world of aquatic eddy correlation, for his valued patience and mentorship, continued encouragement, stimulating conversations, and for many memorable hours in the field. Furthermore I thank my second co-advisor, Prof. Søren Rysgaard, for supporting my research stays in Greenland. These were life-changing experiences for me, and I am proud and happy to have been part of his research group there.

I would like to acknowledge the Faculty of Science at the University of Southern Denmark for providing an excellent platform from which to pursue my Ph.D. studies. I would like to express my deep gratitude to colleagues and friends, too long to list, within the Institute of Biology and in particular in the Nordic Centre for Earth Evolution (NordCEE). Special thanks go to the members of the ‘Glud Lab’, to Dr. Kasper Hancke and to my officemate Dr. Lorenzo Rovelli for many stimulating debates and for being great co-fieldworkers. I must express my gratitude to Anni Glud, Mette Andersen, and Lone Nørgaard Bruun, for their help with all matters administration related.

A heartfelt ‘Qujanaq!’ to the staff and students at the Greenland Climate Research Centre for a very special 16 month research stay in Nuuk. In particular, thank you to Heidi Sørensen, Lorenz Meire, Dr. Martin Blicher, Dr. Thomas Juul-Pedersen, and Peter Schmidt Mikkelsen. I would like to thank the research groups at the Scottish Association for Marine Sciences (Dr. Henrik Stahl) and Heriot-Watt University (Prof. Murray Roberts) in Scotland, and the MPI-AWI research groups in Germany (Dr. Frank Wenzhöfer and Prof. Antje Boetius), who allowed me to join research expeditions they organized. Much of this work is in progress and therefore is not included in this thesis.

Finally I am indebted to those closest to me: my family back home in Malta, who from a time way before starting my Ph.D. degree has been nothing but encouraging and supportive of my (often questionable) endeavours. I am especially indebted to my mother and father, Annette and Joe. This thesis is dedicated to them. And to Lisa, who has always been around to provide reassurance and perspective.

Karl Attard

Odense, September 2014

ABSTRACT

The benthic environments on the continental shelves are hotspots for carbon turnover in the oceans, hosting a high biomass of heterotrophic microbes and fauna as well as benthic primary producers such as seagrasses, micro- and macro-algae, and corals that contribute substantially to ecosystem primary production. While the importance of quantifying the carbon turnover in shelf environments is clear, the global database remains limited. This is especially true for complex benthic habitats that are widespread on the continental shelf such as consolidated sands and gravels, rocky and uneven substrates, and for those dominated by large macroalgae and macrofauna. The difficulties in applying traditional benthic O2 exchange measurement techniques such as benthic microprofiling and benthic chamber incubations have left many shelf environments practically unstudied in the context of organic carbon turnover. Data is particularly scarce for the massive coastal regions of the Arctic that encompass around 25% of the global continental shelf area. Many of these regions are undergoing dramatic transformations due to climate change that are expected to affect coastal productivity.

The primary aim of this thesis was to investigate the organic carbon turnover in complex, understudied benthic environments on the continental shelf using the aquatic O2 eddy correlation (EC) method. In contrast with the traditional measurement techniques, the EC method is not confined to soft sediments, and the measurements integrate a large area of the seabed (typically 10-100 m2) under the natural light and flow conditions. This thesis draws on three case studies. The first study presents a series of high-quality EC measurements from an intensive 13 month study that was carried out to estimate benthic primary production and carbon turnover in a Greenland fjord. The second study investigates the benthic O2 uptake rate of two cold-water coral communities in the North Atlantic using EC. The third study is a seasonal study investigating benthic primary production and carbon turnover in a pristine maerl bed and a nearby sandy habitat in Loch Sween, Scotland.

These high-quality novel data document that hard and complex benthic surfaces mediate high carbon turnover rates. These habitats play a key role in the remineralization of organic matter, regeneration of nutrients, and benthic primary production of temperate and polar coastal waters. The EC method estimates the benthic O2 exchange rates over large areas of the seabed in environments where other techniques would likely fail. Altogether, these studies provide a number of exciting and timely avenues for conducting future research.

DANSK RESUMÉ

De bentiske økosystemer på de kontinentale sokler er brændpunkter for karbon omsætning i havet. De opretholder/indeholder en høj biomasse bestående af heterotrofe mikrober, fauna lige såvel som bentisk primær producenter, så som ålegræs, mikro- og makro alger og koraller. Disse bidrager en betydelig del til hele økosystemets primær produktion. På trods af vigtigheden af at kunne kvantificere karbon omsætningen på de kontinentale sokler, er den globale database fortsat begrænset. Dette er især gældende for komplekse bentiske habitater bestående af konsolideret sand og grus, klippe og ujævne overflader som primært er domineret af makroalger og – fauna, da disse udgør en betragtelig andel af de kontinentale sokler. Problematikken i at benytte traditionelle metoder for måling af bentisk ilt udveksling, så som ilt mikroprofilering og iltkamre, har forårsaget at mange områder fortsat ikke er undersøgt i forbindelse med karbon omsætning. Især data/information fra den arktiske region, som udgør 25% af de globale kontinentale sokler, er manglende. Mange af økosystemerne i denne region gennemgår dramatiske forandringer grundet klima forandring, som forventes at påvirke den kystnære primær produktionen.

Det primære mål med denne afhandling var at undersøge omsætningen at karbon i de komplekse og relative uudforsket bentiske systemer på de kontinentale sokler ved hjælp af EDDY CORRELATION (EC). I kontrast til med de traditionelle måle tekniker, er EC ikke begrænset til blødbund sedimenter og integrerer et større område (typisk 10-100 m2) under naturlige lys- og strømnings forhold. I det første studie blev EC metoden anvendt i overfladiske (3-22 m) og forskellige bentiske habitater i en Grønlands fjord, for at estimere den bentiske primær produktion og karbon omsætning over et år. Studiet viste tilstedeværelsen af et aktiv og produktivt fotosyntetiserende samfund hele året, som er i stand til at fotosyntetisere under meget lave lys forhold. Overordnet, fremhæver disse målinger at betydningen af bentisk primær produktion i Arktis på nuværende tidspunkt kan være understimeret.

Det andet studie undersøger det bentiske ilt optag af et kold-vands koralrev under dannelse ved hjælp af EC. Disse er bemærkelsesværdige langsomt voksende dybhavs habitater, som findes over hele verdenen. Disse strukturelle komplekse rev fungerer som habitat for et diverst og rigt samfund, som forventes at omsætte betragtelige mængder at organisk materiale. Det er dog vanskeligt at benytte traditionelle måle tekniker for disse områder, hvilket betyder at de fortsat er forholdsvis uudforskede. I dette studie, blev EC målinger benyttet ved to koldvandskoralrev lokaliseret i helholdsvis Sea of Hebridget, Skotland (dybde 138 m) og Sjernsund, Norge (220 m). For begge områder viste EC undersøgelserne et iltoptag, som var 4-5

gange større end tidligere reporteret for blødbunds områder fra samme vanddybder. Dette antyder at koldvandskoralrev er vigtige brændpunkter for karbon omsætning i dybhavet.

Det tredje studie undersøger den bentiske ilt udveksling i et uberørt område med koraller i den tempererede fjord Loch Sween, Skotland i en dybde af 5 m. Koralrevet består af fritlevende rødalger, som akkumuler på illuminerede bundområder. Disse langsomt voksende og strukturelle komplekse habitater bestående af rødalger, er almindeligt udbredt og lægger grund for en rig biodiversitet af autotrofe organismer, så vel som makrofauna. Dog er de fortsat overraskende uudforsket. Dette studie viser de første estimater for primær produktion og karbon omsætning for disse koralrev over forskellig sæsoner ved brug af EC. På trods af den betragtelige bentiske primær produktion, viste disse koralrev sig at være net- heterotrofe året rundt. Samtidigt, viste den lineære P-I sammenhæng en undermætning af lys. For samlingens skyld, blev EC målinger ligeledes udført på et sandområde i nærheden (indenfor 20 m) og viste en signifikant forskel på den bentiske ilt dynamik mellem områderne. Parallelle målinger foretaget med bentiske kamre under mørke viste en afvigelse på op til 8 fold mellem kamrene. Men, på trods af den omfattende småskala forskellighed, var middelværdien af de parallelle kammermålinger og EC målingen for begge bentiske habitater indenfor 20 % af hinanden.

Overordnet, viser studierne beskrevet i denne afhandling at de bentiske samfund på hårde- og strukturelle komplekse overflader understøtter en høj karbon omsætning og derfor har en stor betydning for mineraliseringen af organisk materiale, regenereringen af næringssalte og primær produktionen i Tempererede- og Arktiske kystområder. Ydermere, viser disse studier at EC er et værdifuldt redskab til målinger af den bentiske ilt udveksling i komplekse bentiske områder og under varierende miljømæssige forhold.

LIST OF MANUSCRIPTS

Paper I

Title: Seasonal rates of benthic primary production in a Greenland fjord measured by aquatic eddy correlation.

Authors: K. M. Attard, R. N. Glud, D. F. McGinnis, and S. Rysgaard

Journal: Limnology and Oceanography 59: 1555-1569

Status: Published (2014)

Paper II

Title: Benthic O2 uptake of two cold-water coral communities estimated with the non- invasive eddy correlation technique.

Authors: L. Rovelli, K. M. Attard, L. D. Bryant, S. Flögel, H. Stahl, J. M. Roberts, P. Linke, and R. N. Glud

Journal: Marine Ecology Progress Series

Status: Submitted (08/2014)

Paper III

Title: Benthic O2 exchange in a pristine maerl bed and a sandy habitat in a temperate sea loch (Loch Sween, Scotland): A seasonal, in situ study.

Authors: K. M. Attard, H. Stahl, N. Kamenos, G. Turner, H. L. Burdett, and R. N. Glud

Journal: Limnology and Oceanography

Status: In preparation

Introduction The coastal marine carbon cycle

1. INTRODUCTION

Preface

The purpose of the forthcoming sections is to provide a brief, broad introduction to near

shore benthic environments as hotspots for carbon turnover within the oceans. The

different compartments of the benthic system are presented within the context of the

coastal carbon cycle: benthic primary production and the attenuation of light in the water,

organic carbon degradation, and the importance of benthic fauna. This is followed by a

discussion on the benthic O2 exchange rate as a proxy for benthic carbon mineralization

and benthic primary production rates. The introduction concludes with a description of the

most popular methods used to estimate the benthic O2 exchange rate, outlining their

strengths and weaknesses, with a special focus on the aquatic O2 eddy correlation method

as the main approach used in this thesis.

1.1. THE COASTAL MARINE CARBON CYCLE

The oceans cover more than 3.6 x 108 km2, or approximately 70% of Earth’s surface, and

have an average depth of 3.6 km (Charette & Smith 2010). In the simplest terms, the global

oceanic biosphere is contained within two distinct compartments: the water column,

termed the ‘pelagic’ zone, and the underlying seabed, termed the ‘benthic’ zone. The

resultant volume of 1.3 x 109 km3 exceeds by far that of the most biologically active layer of

the seabed that is restricted to the upper metres of the seabed sediments (Glud 2008). The

space available for organic matter synthesis and degradation to take place is therefore

greatest within the water column; however, the volume-specific rates of organic matter

production and degradation within the surface sediments typically are 100-1000 times

Introduction The coastal marine carbon cycle

higher (Glud 2008). This is especially the case for benthic environments on the continental

shelves, typically having depths less than 200 m. Continental shelves receive up to 50 % of

the pelagic primary production through sedimentation (Wollast 1991), harbour a

substantial biomass of benthic primary producers (Cahoon 1999), and release regenerated

nutrients such as nitrogen, phosphorus, and silica that can supply up to 80% of

phytoplankton nutrient requirements (Middelburg & Soetaert 2004). Sedimentation and

benthic primary production is closely coupled with high densities of heterotrophic microbes

and fauna that efficiently recycle nutrients and carbon. Altogether, continental shelves are

responsible for ~40% of the global oceanic benthic carbon mineralization (Glud 2008), and

when considering that they encompass an area of just 2.7 x 107 km2, or 7.5% of the ocean

area, the importance of studying these benthic environments becomes clear.

Benthic primary production

Benthic primary producers such as seagrasses, micro- and macro-algae, and corals increase

habitat complexity and biodiversity, serve as nursery grounds for juvenile invertebrates and

fish, are hotspots for carbon burial within the ocean, and contribute directly to ecosystem

primary production (Duarte et al. 2005). Indeed, one compelling reason to investigate the

carbon cycle at the seabed in coastal environments is to quantify the benthic primary

production. Large regions of the coastal seabed receive a substantial portion of the down

welling sunlight, and especially in clear oligotrophic waters of e.g., the Arctic and the

tropics, benthic photosynthesis typically dominates coastal primary production (Glud et al.

2009). Benthic phototrophic communities are in a favourable location to assimilate the

regenerated nutrients that are released from the efficient organic matter mineralization at

the seabed, and effectively strip out the nutrients and outcompete the pelagic

phytoplankton. Therefore, while nutrients limit phytoplankton growth in many regions of

Introduction The coastal marine carbon cycle

the sunlit oceans, the availability of light is considered the main factor determining the

global distribution of benthic phototrophic communities (Gattuso et al. 2006).

Light availability

Like other photosynthetic organisms, benthic primary producers utilize the spectral range

of solar radiation during photosynthesis from 400 to 700 nm, termed ‘photosynthetically

active radiation’ (PAR). Sunlight reaching the ocean surface is attenuated exponentially

with depth as a consequence of scattering and absorption by water molecules,

phytoplankton, suspended particulate inorganic material, and coloured dissolved organic

matter (cDOM) (IOCCG 2000). The amount of PAR reaching the seabed can therefore vary

substantially due to e.g., outflow of streams and rivers, wave- and current-generated

resuspension of bottom sediments, and phytoplankton blooms. Furthermore, the intensity

of the downwelling light and the duration of the daytime period vary seasonally, and in the

Polar Regions this variation is particularly extreme.

The depth limit of primary production in the ocean is defined by the ‘compensation depth’,

the depth at which the photosynthetic organisms receive enough light for gross

photosynthetic carbon fixation to balance respiratory losses over the course of one day. For

a given system, the average compensation depth is typically taken as the depth

corresponding to 1% of the PAR at the surface (Falkowski & Raven 2006). However, the

light requirements of photosynthetic organisms vary greatly. For example, live benthic

diatoms have been collected from depths up to 191 m, corresponding to 0.028% of surface

PAR (McGee et al. 2008). Similar observations were made for leathery, foliose, and crustose

macroalgae, the latter occurring at depths exceeding 250 m, or 0.0005% of surface PAR

(reviewed by Markager & Sand-Jensen 1992). Clearly, there is potential for a substantial

benthic primary production on the continental shelves. Model studies indicate that a net

Introduction The coastal marine carbon cycle

benthic photosynthetic production can occur over 33% of the global shelf area (Gattuso et

al. 2006). However, the global database on benthic primary production remains limited,

especially for the Arctic, and the importance of this component to ecosystem production

could be underestimated (Cahoon 1999; Glud et al. 2009; Paper I). A better quantitative

understanding of benthic primary production is imperative if we are to constrain the

carbon and nutrient budgets as well as identify the responses of coastal systems to e.g.,

eutrophication and climate change. In this context it is also highly relevant to quantify the

benthic remineralization rate of organic matter, a process that is intimately coupled with

benthic primary production.

Organic carbon degradation

In coastal benthic environments, the organic carbon degradation is largely mediated by

microbes utilizing various pathways of oxidation: aerobic respiration using oxygen (O2) as

an electron acceptor is the most thermodynamically favourable of the abundant species,

but it is rapidly depleted within the top millimetres or centimetres of the seabed sediment.

Following hydrolysis of the large particulate polymers into smaller molecules that can be

transported across the cell membranes, the carbon that has not been degraded aerobically

is degraded by a mutualistic consortium of anaerobic heterotrophic bacteria that utilize

inorganic compounds such as nitrate, manganese oxides, iron oxides, and sulphate as

electron acceptors (Canfield et al. 1993). The importance of the individual respiration

pathways for the total carbon turnover in coastal sediments depends on the availability of

the inorganic compounds and the energetics (i.e., energy yield) of the individual reactions

(Canfield et al. 2005). A good example of the interplay between energetics and availability

is the importance of sulphate reduction to the total carbon turnover in many coastal

Introduction The coastal marine carbon cycle

benthic settings, that typically averages 30-50%, even though the energy yield from this

process is ~8 fold lower than those via nitrate and manganese oxide (Canfield et al. 2005).

Typically, microbial O2 respiration constitutes less than 50% of the total carbon

mineralization in shallow waters, so the benthic carbon degradation is largely mediated

anaerobically (Canfield et al. 2005). However the reduced products from anaerobic

degradation are to a large extent reoxidized by O2, either directly or via a series of complex

abiotic and microbial catalysed redox processes termed the ‘redox cascade’ (Fenchel &

Jørgensen 1977; Fig. 1). Thus, from a microbial benthic carbon turnover perspective, O2 is

used for both the aerobic heterotrophic respiration, as well as for the reoxidation of

reduced inorganic compounds that are produced during anaerobic carbon mineralization.

As a result, the rate at which O2 is taken up at the seabed, termed the ‘benthic O2 uptake

rate’, has established itself over the years as the most widely used proxy for the total

microbial benthic carbon mineralization (Glud 2008). When large areas of the seabed are

integrated in the benthic O2 uptake measurements, the measurements also include

contributions from benthic fauna, a ubiquitous and important component of near shore

habitats.

Importance of benthic fauna

Suspension-feeding benthic fauna such as polychaetes, bivalves, and echinoderms

frequently occur in very high densities (several thousand ind. m-2) and are known to

enhance the pelagic-benthic coupling in coastal environments by actively trapping

particulate matter suspended in the water column (Dame 1996; Davoult & Gounin 1995;

Thomson & Schaffer 2001). On an annual basis, the carbon demand of the dominant

species of benthic fauna may constitute a substantial percentage of the coastal primary

Introduction The coastal marine carbon cycle

production (Thomson & Schaffner 2001; Blicher et al. 2009). Therefore, the benthic fauna

are an important component of the coastal carbon cycle.

Benthic fauna such as sediment-dwelling polychaetes and brittle stars markedly alter the

surface sediments through bioturbation, a term that encapsulates both the processes of

sediment reworking (i.e., movement of sediment particles) as well as sediment or burrow

ventilation (Kristensen et al. 2012). The renewal of burrow water through irrigation

increases the oxic volume of the sediment, stimulates aerobic microbial activity along the

burrow walls, and enhances the chemical oxidation of reduced products of anaerobic decay

by introducing O2 in an otherwise anoxic environment (Kristensen & Kostka 2005;

Jørgensen et al. 2005). Bioturbation increases the exposure time of refractory organic

materials such as lignin to oxic conditions under which these compounds can be degraded

(Hulthe et al. 1998; Kristensen & Holmer 2001; Emerson & Hedges 2003). The complex

redox patterns surrounding oxic microenvironments are thought to yield faster and more

complete microbial decomposition of compounds such as lipids over short timescales (days

to months) (Sun et al. 1999). However, over long (geologic) timescales, dominant anaerobic

pathways oxidize equal amounts of organic matter as oxic respiration under typical near

shore sedimentation rates that exceed ~0.1 cm y-1 (Canfield 1989).

Introduction The coastal marine carbon cycle

Fig. 1: In the absence of benthic photosynthesis, the benthic O2 uptake rate integrates the aerobic respiration of fauna and microbes and the reoxidation of reduced products from anaerobic decay. Green arrows represent respiration, red arrows oxidation, and the orange arrows indicate minor electron sinks not

included in the O2 uptake measurements such as N2 release from denitrification

and burial of authigenic minerals such as pyrite (FeS2). After Glud (2008).

Introduction The benthic O2 exchange rate

1.2. THE BENTHIC O2 EXCHANGE RATE

Benthic O2 exchange rates as a measure of C remineralisation

In the absence of benthic photosynthesis, the rate of O2 exchange between the seabed and

the overlying water column integrates (1) the aerobic heterotrophic activity of microbes

and fauna, and (2) the reoxidation of inorganic products that are released following carbon

degradation via anaerobic pathways. The relative importance of each of these components

to the benthic O2 uptake rate by the sediment can vary substantially and depends on the

environmental setting. In general, the faunal contribution to the benthic O2 uptake is

typically around 50% for shallow water benthic environments, of which less than half of

that 50% constitutes faunal respiration. Rather, the majority of the faunal contributions are

due to bioturbation that stimulates both the microbial aerobic respiration as well as

chemical reoxidation (e.g., Jørgensen et al. 2005). Altogether, reoxidation typically

accounts for more than 50% of the benthic O2 uptake rate in shallow ecosystems, with the

remaining percentage being partitioned between aerobic respiration by microbes and

fauna (Glud 2008; Fig. 1). On an oceanic scale, the faunal contribution to the benthic O2

uptake decreases with depth until the microbial-mediated O2 uptake completely dominates

at depths greater than 3500 m (Glud 2008).

The conversion of benthic O2 uptake rates into depth-integrated benthic C mineralization

rates invokes several assumptions. Firstly, the benthic O2 uptake rate does not include

electron sinks such as N2 release from denitrification, the O2 equivalents used during

nitrification, and burial of reduced diagenetic products such as FeS2. While the individual

processes can be quantified independently through additional analyses, altogether they

typically account for only a small fraction (<20%) of the electron equivalents relative to the

total carbon remineralisation process, and are often assumed to cancel each other out

Introduction The benthic O2 exchange rate

(Canfield et al. 2005; Glud 2008). The second assumption is that the production and

oxidation rate of reduced products of anaerobic decay is at steady state. Detailed studies

have shown a decoupling between production and oxidation, resulting in storage of

reduced compounds that accumulate over both short (diel) and long (seasonal) timescales.

For example, anaerobic metabolites that accumulate during darkness create an ‘O2 debt’

that is repaid during periods of oxygenation. Consequently, during periods affected by

storage, the sediment-water O2 exchange underestimates both the O2 efflux due to primary

production during the daytime as well as O2 uptake due to C remineralisation during night

(Fenchel & Glud 2000). To account for the storage effect, the approach typically has been

to concurrently measure the exchange of dissolved inorganic carbon (DIC) alongside O2.

The DIC exchange represents a more direct measure of the depth-integrated benthic C

remineralisation rate since DIC constitutes an end product of each of the C remineralisation

reaction pathways (Canfield et al. 2005). Thus, the ratio between the DIC and O2 exchange

rates has been used to evaluate the balance between production and oxidation of reduced

solutes (e.g., Therkildsen & Lomstein 1993). Ratios of up to ~5 have been reported for

shallow-water sediments; however, the DIC:O2 typically ranges seasonally from 0.8 to 1.2

and approaches 1.0 when integrated annually (Glud 2008). While it is advisable to quantify

both the DIC and O2 exchange rates when estimating the benthic C remineralisation, for

various reasons it is not always possible to collect and/or analyse DIC samples in the field.

In contrast, O2 is readily measurable, and therefore the benthic O2 exchange rate remains

the most widely used approach for estimating the total benthic C mineralization rates (Glud

2008).

Introduction The benthic O2 exchange rate

Estimating the benthic primary productivity from O2 exchange measurements

Benthic primary producers such as benthic diatoms, seagrasses, and kelps produce O2

during the daytime as a by-product of photosynthesis. At the seabed, autotrophic O2

production is superimposed onto the O2 consumption driven by heterotrophic microbes,

fauna, and chemical reoxidation processes. During daytime, the benthic O2 exchange rate is

a measure of the balance between production and consumption of O2, and is termed the

‘net primary productivity’ (NPP). During dark, the NPP is driven only by the O2 consumption

and is termed the nighttime O2 respiration (R). ‘Respiration’ here encapsulates aerobic

respiration as well as the O2 equivalents used for reoxidation. The sum of R with NPP in the

light is the gross primary productivity (GPP) (Fig. 2). In practice, the process of quantifying

the R during daytime is not straightforward, and the approach of estimating the GPP from

O2 exchange measurements has generally been to assume a light-independent rate of R

(Glud et al. 2009). However, R during light is typically ~2 fold higher than R during dark due

to (1) deeper O2 penetration during daytime (and hence a larger volume of sediment that

can support aerobic remineralisation), and (2) leaching of labile photosynthesates that

stimulate microbial turnover (Epping & Jørgensen 1996; Fenchel & Glud 2000).

Furthermore, infauna of shallow water sediments frequently exhibit oscillations in their

activities in response to environmental cues such as the light-dark cycle. If in high enough

densities, the integrated activity of the organisms can substantially increase the benthic O2

consumption rate during periods of activity (Wenzöfer & Glud 2004). Therefore, while the

assumption of a light-independent R rate is practical and has been widely adopted

especially in field studies, it typically underestimates the daytime R rate and subsequently

also the derived GPP rates.

Introduction The benthic O2 exchange rate

Fig. 2: Schematic diagrams illustrating the differences between gross primary productivity (GPP) and net primary productivity (NPP) for a typical shallow-water benthic environment. For NPP, when PAR >0, respiration is masked by a photosynthetic O production. GPP is then the sum of the respiratory losses (R) with 2 NPP, assuming a light-independent rate of R (see text). Positive values indicate a

release of O2, negative values indicate an uptake.

In addition to NPP, GPP, and R, the benthic O2 exchange rate is also used to derive a fourth

parameter termed the ‘net ecosystem metabolism’ (NEM). The NEM indicates whether

sediment O2 production through photosynthesis balances the various heterotrophic

processes that directly or indirectly consume O2 over a 24 h period, and it is derived as the

average of NPP and R rates weighted to the number of daylight and nighttime hours,

respectively. Positive NEM values indicate a net O2 release by the benthic ecosystem

(autotrophy) while negative NEM values indicate a net O2 uptake (heterotrophy) over a 24

h period. For example, this analysis is useful to investigate to what extent the benthic

communities rely on external inputs of organic matter from the pelagic component, and

whether the benthic communities are a net source or sink of O2 and CO2 to shallow coastal

waters (Dalsgaard 2003; Papers I and III).

Evaluating the benthic community light response

The light dependency of the benthic O2 exchange rate is typically investigated using the

photosynthesis vs. irradiance (P-I) relationship. Here, the benthic O2 exchange rates are

Introduction The benthic O2 exchange rate

plotted against the in situ PAR data and the relationship between the two is evaluated

using simple curve-fitting models termed ‘light-saturation curves’ (Jassby & Platt 1976).

One of the most widely used models for evaluating the light response of benthic

communities is the tangential hyperbolic function that is offset by a respiration term i.e.,

NPP = Pm tanh(I/Ik) – R, where Pm is the maximum rate of GPP, I is the in situ PAR (in µmol

-2 -1 -2 -1 quanta m s ), Ik is the light saturation parameter (in µmol quanta m s ), and R is the rate

of O2 uptake during dark (Fig. 3). For benthic systems, the NPP, GPP, and R rates are

typically presented as depth-integrated areal O2 exchange per unit time, in units of mmol

-2 -1 -2 -1 O2 m d or mmol O2 m h . Subsequently, the light level at which photosynthetic O2

production balances sediment O2 consumption, termed the ‘compensation irradiance’ (Ic, in

µmol quanta m-2 s-1), is computed as the x-intercept (Fig. 3). It is sometimes the case that a

net light saturation of the benthic O2 exchange rates is not observed under field conditions,

and thus the P-I relationship can be better explained (in terms of the R2 value) by a linear

regression (e.g., Jahnke et al. 2008; Rheuban et al. 2014; Paper III).

Traditionally, P-I relationships have been used to assess the photosynthetic performances

of isolated phototrophic communities of e.g., phytoplankton and intact benthic microalgal

mats (Falkowski & Raven 2006; Hancke et al. 2014). However, the same relationship has

been shown to hold remarkably well for mixed benthic communities such as seagrass beds

(Rheuban et al. 2014), permeable sediments (Jahnke et al. 2008; Berg et al. 2013; Papers I

and III), tropical coral reefs (Long et al. 2013), and maerl beds (Martin et al. 2007; Paper III).

As outlined in the above sections, benthic O2 exchange measurements integrate many

different processes, and therefore this should be considered when interpreting the

parameters that are derived from the P-I relationships.

Introduction The benthic O2 exchange rate

Fig. 3: A P-I relationship (solid red curve) as fitted to the example NPP dataset presented in Fig. 2. The main parameters that are typically derived from a P-I curve fitting are outlined, and broken lines are used to illustrate the meaning of each parameter.

Introduction Estimating the benthic O2 exchange rate

1.3. ESTIMATING THE BENTHIC O2 EXCHANGE RATE

The benthic boundary layer

The benthic O2 exchange rate is estimated from measurements at or close to the interface

between the seabed and the overlying water column, from within a region termed the

benthic boundary layer (BBL). The BBL is generated by the friction of the water movement

over the seabed and is typically classed into three zones that are defined in terms of the

dominant momentum and solute transport processes. The ‘logarithmic layer’ occupies a

large portion of the BBL volume and here transport is largely mediated by turbulent eddy

diffusion at a rate that may exceed molecular diffusion by several orders of magnitude

(Wüest & Lorke 2003). Approaching the seabed the vertical motions of turbulence are

suppressed until viscous forces start to dominate at a height of around 10 mm above the

seabed surface. This region is termed the ‘viscous sublayer’ and is the region where most of

the changes in flow velocity take place (Dade et al. 2001). Closest to the seabed, turbulent

and viscous forces are dampened to the extent that molecular diffusion dominates the

solute transport in a region of ~1 mm or less thick termed the ‘diffusive boundary layer’

(DBL) (Boudreau 2001; Fig. 4). Therefore, the DBL may become a ‘bottleneck’ for transport

of O2 between the seabed and the overlying water column if the production or uptake rate

of O2 by the sediment exceeds the maximal transport rate of O2 through the DBL (Kühl et al.

1996; Lorke et al. 2003; Glud et al. 2007). The exception to this is for highly permeable

sediments, typically defined as having permeability greater than 10-12 m2, where advective

pore-water flow dominates the sediment-water exchange of O2 and therefore bypasses the

restrictions imposed by molecular diffusion (Huettel & Gust 1992).

Introduction Estimating the benthic O2 exchange rate

Fig. 4: A conceptual illustration of the BBL. Note that the scale on the left hand side is non-linear. The dominant transport processes within each layer are

outlined: turbulent eddy diffusion (Kρ) in the lograthmic layer, molecular viscosity (υ) in the viscous sublayer, and molecular diffusion (D) in the DBL.

Benthic O2 microprofile and chamber measurements

The two commonest procedures for quantifying the benthic O2 exchange rate are sediment

O2 microprofiling and sediment O2 incubation approaches (Glud 2008; Glud et al. 2009; Fig.

4). Sediment O2 microprofiles are obtained from microscale O2 concentration

measurements across the sediment-water interface using O2 microsensors with a tip

diameter typically between 10-50 µm. From these measurements, the ‘diffusive O2

exchange rate’ (DOE) is computed from either the concentration gradient within the

diffusive boundary layer (DBL) or from below the sediment surface, in each case applying

the relevant transport coefficient for O (D), as ( ) (termed ‘Fick’s first law’ of 2

diffusion). This analysis quantifies the net efflux or uptake of O2 by the sediment surface at

a single point and therefore excludes the faunal (respiration and bioturbation) and

macroalgal contributions. The curvature of the O2 microprofiles may be used to compute

Introduction Estimating the benthic O2 exchange rate

the volume-specific rates (DOEvol) of the sediment layers using Fick’s second law of

diffusion as (Berg et al. 1998), thereby revealing the internal sediment

structure in O2 production and consumption on spatial scales that are relevant to microbial

processes. This allows for detailed measurements on microbial-mediated sediment O2

dynamics such as O2 respiration, net and gross photosynthesis, and photosynthesis-coupled

respiration (Kühl et al. 1996). For abyssal and hadal sediments, the DOE gives a good

approximation of the total O2 uptake by the benthos, since at these depths the

contribution of benthic fauna to the benthic O2 uptake is small (Glud 2008). However in

coastal environments the faunal contribution may constitute 50% or more of the total

benthic O2 uptake rate, and therefore a larger spatial resolution that integrates the faunal

contributions is required. Oxygen microsensors are very fragile and therefore

measurements are restricted to soft sediments.

Fig. 5: Well-established methods for estimating the benthic O 2 exchange rate include microprofiling techniques (A) that may be carried out in the laboratory or in situ, and enclosure techniques (incubation) of sediment and overlying water in recovered sediment cores (B) or in situ using benthic chambers (C). Adapted from Glud (2008). 32

Introduction Estimating the benthic O2 exchange rate

Benthic O2 exchange rates of mixed benthic communities have traditionally been estimated

using enclosure techniques. Here, a known sediment area (a) and overlying water volume

(v) is enclosed and the benthic O2 exchange (BOE) is computed from the rate of increase

(production) or decrease (consumption) in O2 concentration (C) in the well-mixed overlying

water phase over time (t) as BOE = (v/a)*(∂C/∂t). This analysis may be carried out both in

the laboratory using recovered sediment cores or in situ using benthic O2 chambers (Fig. 5).

Enclosure techniques are well-established and have been widely employed in both coastal

as well as deep-sea systems. Exchange rates of O2 are derived from a well-defined sediment

area, and allow for estimation of the exchange rates of other solutes of interest such as

nutrients and DIC. Incubations are frequently combined with microprofile measurements,

and the difference between the chamber-resolved and microprofile-resolved O2 uptake

rates are attributed to the faunal contributions (e.g., Glud et al. 2003). Furthermore, post-

incubation analyses on the incubated sediment such as quantifying the phototrophic and

faunal biomass provide useful additional information and complement the O2 exchange

measurements (e.g., Dalsgaard 2003).

Typically, incubations of near shore sediments last only few hours and even large benthic

chambers have a spatial resolution of less than 0.2 m2, and therefore deployment of

multiple chambers in parallel is required, to increase the spatial resolution and obtain more

representative estimates of the benthic O2 exchange rate (Glud & Blackburn 2002; Paper

III). Perhaps the most substantial shortcoming of the incubation approach is that the

method requires insertion of the chamber or core liners into the seabed and thus, like the

microprofiling method, it is largely restricted to soft sediments.

The enclosure process excludes the natural hydrodynamics and the rotating stirrer defines

a turbulence regime within the chamber that is markedly different from that in situ (e.g.,

Introduction Estimating the benthic O2 exchange rate

Férron et al. 2008). The hydrodynamics may influence various components of the benthic

ecosystem. For example, high flow velocity at the seabed can enhance rates of benthic

primary production by (1) facilitating the efflux of O2 from the phototrophic organisms and

thus increasing the affinity of RuBisCO to CO2 (Mass et al. 2010), and (2) by facilitating

influx of nutrients and DIC required for photosynthesis (Enriquez & Rodriguez-Roman 2006;

Cook & Røy 2006). For highly permeable sediments, a tight coupling exists between the

benthic O2 uptake rate and the flow velocity. Increased flow stimulates advective flushing

of surface sediments with oxygenated water, increases the benthic O2 uptake rate (Cook et

al. 2007; Berg et al. 2013), and promotes the decomposition of dissolved organic carbon

(DOC) (Chipman et al. 2010). The hydrodynamics also influences the feeding behaviour of

benthic fauna such as suspension-feeding brittle stars by increasing particle encounter

rates through increased turbulence and resuspension of settled material (Davoult & Gounin

1995). Therefore, the best estimates of the benthic O2 exchange rate in coastal waters

would need to integrate the natural hydrodynamics at the seabed.

The aquatic O2 eddy correlation (EC) method

More recently, the aquatic O2 eddy correlation (EC) method by Berg et al. (2003) was

introduced that allows estimating the benthic O2 exchange rate non-invasively and over

large (~10-100 m2) areas of the seabed. While the region that contributes most to the flux

is smaller than the overall EC footprint size (the footprint size is typically defined as the

smallest area that contributes to 90% of the flux; Berg et al. 2007), the EC method still has a

much larger spatial resolution than benthic chambers and can therefore include the

contributions from large benthic fauna and macroalgae (Fig. 6). The EC measurements are

carried out away from the seabed surface so the method is not confined to soft sediments.

Indeed, this method has been applied to complex benthic surfaces such as rocky

Introduction Estimating the benthic O2 exchange rate

embayments (Glud et al. 2010), seagrass beds (Hume et al. 2011; Rheuban et al. 2014),

permeable sediments (Reimers et al. 2012; Berg et al. 2013; McGinnis et al. 2014), oyster

beds (Reidenbach et al. 2013), and tropical coral reefs (Long et al. 2013). Furthermore, the

EC method has also been applied to quantify the O2 exchange at the sea ice-water interface

(Long et al. 2012; Glud et al. 2014), and holds great potential as a tool to investigate the

activity at this highly complex and dynamic interface.

Fig. 6: EC measurements integrate large areas of the seabed and can include contributions from large macrofauna that are difficult to capture using benthic chambers (A; Berg et al. 2003). The footprint area (B) is defined as the smallest area that contributes with 90% of the flux. Its characteristics are computed from dissolved conservative tracer release simulations, and are dependent upon the measurement height and the sediment surface roughness (Berg et al. 2007).

The EC method combines high-frequency Eulerian measurements of the flow velocity (u, v,

w) and the O2 concentration taken at a single point located typically 10-30 cm above the

seabed surface within the turbulent BBL. The basic principles of EC O2 measurements are

based on the covariance of vertical velocity with deviatory O2 concentration. Over time, this

gives a net O2 transport towards or away from the sediment surface, depending on whether

the benthos is consuming or releasing O2 (Fig. 7). Through ‘Reynolds decomposition’ theory,

the measured vertical velocity (w) and the O2 concentration (C) are decomposed into a

mean and instantaneous value as 〈 〉 and 〈 〉 , where the angle

Introduction Estimating the benthic O2 exchange rate

brackets denotes time averaging. Subsequently the O2 fluxes are derived as the covariance

〈 〉 (Berg et al. 2003).

Fig. 7: Illustrations of the basic measurement principles of the EC method. The upper panel shows a conceptual model for an eddy that moves parcels of water (C1 or C2)

with speed w1 or w2. By knowing the O2 concentration and the speed, one can calculate the flux (adapted from Burba 2013). The lower panel shows an example of

measured vertical velocity (w) and O2 concentration over 120 s in a shallow site during daytime in Greenland. The dotted lines are the mean. When w is positive the

O2 concentration is above the mean, and vice-versa, suggesting a net O2 efflux from the sediment surface. During nighttime the opposite would be true i.e., above

average w would coincide with below average O2 concentrations.

Introduction Estimating the benthic O2 exchange rate

Deriving the O2 flux estimates from the raw EC data involves a multiple step process that

has been detailed elsewhere (e.g., Lorrai et al. 2010; Paper I). Many of these routines have

been adopted from the well-established atmospheric EC literature (Lee et al. 2004). In

short, the first series of procedures include: quality-checking the raw EC dataset, averaging,

coordinate rotation of the velocity field, calibration of the raw O2 microsensor signal, and

extracting the O2 fluxes. The derived O2 fluxes are then evaluated for their quality based on

a set of defined criteria, such as exclusion of O2 fluxes during periods of rapid changes in

flow direction, O2 concentration, or flow velocity, or for periods with insufficient turbulent

mixing (Brand et al. 2008; Paper I). Once extracted and quality-checked, the EC O2 fluxes

are plotted as a time series. It is customary to bin the derived EC fluxes into longer time

intervals of 1 or more hours, to reduce the short-term variability due to e.g., non-steady

state conditions within the BBL (Holtappels et al. 2013), and therefore present rates that

are more representative of those at the sediment-water interface. Measurements in

shallow-water environments typically cover one or more days, to help elucidate the

dynamics and drivers of the O2 exchange rate under changing environmental variables such

as flow velocity and PAR (Fig. 8).

Introduction Estimating the benthic O2 exchange rate

Fig. 8: Top panel: An EC O2 time series over 5 days of continuous deployment in a 5 m deep benthic habitat in a Greenland fjord. Bottom panel: drivers of EC O2 exchange rates for the 5-day dataset: light (PAR) availability, and water flow velocity. Data from Paper I.

Introduction Estimating the benthic O2 exchange rate

The main components of an EC O2 system consist of an acoustic Doppler velocimeter (ADV)

and a fast responding O2 microsensor. The ADV most commonly employed for EC

measurements is the 6 MHz Vector (Nortek A/S, Norway) that typically maintains good

signal-to-noise ratios even in clear waters. Most EC studies to date have used

electrochemical Clark-type O2 microsensors with a 10-20 µm tip diameter and a 90%

response time < 0.5 s (Revsbech 1989) that relays the signal to the ADV via a submersible

amplifier (Berg et al. 2003; McGinnis et al. 2011). More recently, optical sensors have been

employed for EC studies that have the advantages of being more robust, are less prone to

signal drift, and are not stirring sensitive (Chipman et al. 2012). Altogether the EC

instrumentation is rather light and self-contained. For shallow-water deployments the

equipment is typically mounted onto a small tripod frame that is designed to minimize

hydrodynamic interferences and can be easily deployed from a small research vessel (e.g.,

Berg & Huettel 2008; Paper I). For deeper deployments, EC instruments have been

mounted onto autonomous benthic landers (Berg et al. 2009) or have had frames

specifically designed for deployment by remotely operated vehicle (ROV; McGinnis et al.

2011, 2014; Paper II; Fig. 9).

Introduction Estimating the benthic O2 exchange rate

Fig. 9: The various EC configurations for benthic studies. Top left: The typical EC configuration for shallow-water deployments, showing the location of the footprint upstream of the instrument (source: L. Cole & P. Berg). Top right: EC setup for measurements in high-energy environments (Reimers et al. 2012). Bottom left: EC system mounted onto an autonomous deep-sea benthic lander (Berg et al. 2009). Bottom right: EC system mounted onto an ROV-deployable frame (McGinnis et al. 2011; Paper II).

Summary of manuscripts

1.4. SUMMARY OF MANUSCRIPTS

The main objective of this thesis was to estimate the benthic O2 exchange rate in complex,

understudied benthic habitats on the continental shelf using the EC method. In the first

study (Paper I), the EC method was applied in different shallow (3-22 m depth) benthic

habitats in a Greenland fjord to estimate the community benthic primary production and

carbon turnover throughout the year. Benthic primary production measurements for the

Arctic remain scarce (Glud et al. 2009), and there only exists one other benthic EC study for

this region that conducted measurements during a single month in May 2009 (Glud et al.

2010). In Paper I, observations are extended over the year to account for the strong

seasonal variations in e.g., light and temperature. The measurements document the

presence of a year round active and productive benthic phototrophic community that is

able to photosynthesize efficiently under very low light levels. This is an interesting and

important finding that highlights the value of conducting seasonal measurements in the

Arctic. Among other things, these measurements indicate that the importance of benthic

primary production in this region could currently be underestimated. A comprehensive

literature review by Cahoon (1999) presented some of the first global estimates for benthic

primary production. At the time of the review, the database for the entire Arctic region

consisted of just two studies. The review suggested a latitudinal-related gradient in benthic

primary production, with the highest rates of production at the tropics and the lowest at

the Poles. However, a recent review by Glud et al. (2009) considered a broader database

for the Arctic, and from this the authors derived an average rate 5-10 times higher than

that presented by Cahoon (1999). Indeed, the results from Paper I are consistent with this

observation, suggesting that the discrepancies between temperate and polar rates of

benthic primary production as reported by Cahoon (1999) could be due to under sampling

rather than due to latitudinal-related gradients (Fig. 10). Furthermore, given the potential

Summary of manuscripts

for benthic primary production in the extensive shelf areas of the Arctic (Gattuso et al.

2006), benthic primary production could become quantitatively more important for

sustaining shallow water food webs under future scenarios of reduced sea ice cover (Glud

et al. 2009).

Fig. 10: Rates for benthic primary production presented in the extensive review by Cahoon (1999). The numbers indicate the number of studies used to derive the average. Note that at the time there were only 2 studies on benthic microalgal production in the Arctic. The original estimates indicate a decrease in benthic primary production from the tropics to the poles. More recent estimates for the Arctic by Glud et al. (2009, 2010) and the seasonal rates in Paper I suggest that the rates presented in Cahoon (1999) are most likely underestimated by 5-10 times. Differences in rates between temperate and polar environments as presented in Cahoon (1999) could therefore be due to under sampling rather than true latitudinal-related gradients.

The second study investigates the benthic O2 uptake rate of reef-forming cold-water coral

(CWC) communities using EC. These are remarkable slow-growing deep-water habitats that

occur worldwide. The complex structural reef framework hosts a great biomass and

diversity of organisms that are expected to turn over considerable amounts of organic

materials (Fig. 11); however, traditional measurement techniques are difficult to apply to

Summary of manuscripts

such environments, and thus they are largely understudied. In this study, EC instruments

were carefully deployed by ROV at two CWC reef sites, one located in the Sea of Hebrides,

Scotland, at a depth of 138 m, and the other located in Stjernsund, Norway, at a depth of

220 m. At both sites, the EC measurements documented rates of O2 uptake by the CWC

-2 -1 communities of ~25 mmol O2 m d that are comparable to shallow water temperate

benthic settings, and four to five times higher than the global mean for soft sediment

communities at comparable depths. This suggests that CWC communities are hotspots for

carbon cycling in deep waters.

Fig. 11: The remarkable cold-water coral habitats investigated. Top panel: the EC instrument deployed at 138 m at Mingulay Reef. Bottom panel: the cold-water corals and associated fauna in Stjernsund, Northern Norway, at a depth of 220 m.

Summary of manuscripts

The third study investigates the benthic O2 exchange rate of a pristine maerl bed

community in temperate Loch Sween, Scotland, at a depth of 5 m. Maerl beds are

composed of dense aggregations of free-living red coralline algae that accumulate on

photic seabed surfaces. They are widespread, slow-growing, structurally complex perennial

habitats that support a very rich biodiversity of autotrophs as well as macrofauna (Fig. 12),

and are surprisingly understudied. This study presents the first rate estimates of maerl

community primary production and carbon turnover using EC and in different seasons.

Despite a substantial benthic primary production, the maerl community was net

heterotrophic year round, and linear P-I relationships indicated a light under saturation. EC

measurements at the maerl bed were compared to measurements carried out at a nearby

(within 20 m) sandy benthic habitat. The EC measurements document substantial

differences in the resolved benthic O2 dynamics between the two sites. Benthic chambers

deployed during dark in parallel with the EC instruments indicated variations of up to ~8

fold between replicate chambers. Despite extensive small scale variability, mean rates of O2

uptake as resolved in parallel by chambers and EC in both benthic habitats were within 20%

of one another. These results are reviewed within the context of the existing literature

comparing benthic chambers and EC measurements in different benthic settings.

Altogether, this thesis presents novel and important insights into the carbon turnover in

hard and complex benthic environments. Even though many of these habitats are

acknowledged as being of major importance to sustain marine biodiversity, little is known

about the basic ecosystem functioning and what governs the distribution of the dominant

organisms. Therefore it is highly relevant to describe and further investigate these systems.

In the future, EC measurements combined with complementary analyses on benthic

organisms and the light distribution will provide valuable knowledge on the functioning of

coastal ecosystems.

Summary of manuscripts

Fig. 12: The pristine maerl beds in Loch Sween were the subject of study for Paper III. Maerl beds are structurally complex and provide many niches for both fauna and flora. The image in the top panel shows the high densities of suspension-feeding brittle stars that anchor themselves to the maerl thalli. Photos by Rob Cook, Heriot-Watt University.

References

1.5. REFERENCES

Berg, P., H. Roy, F. Janssen, V. Meyer, B. B. Jorgensen, M. Huettel, and D. de Beer. 2003. Oxygen uptake by aquatic sediments measured with a novel non-invasive eddy correlation technique. Mar. Ecol. Prog. Ser. 261: 75-833.

Berg, P., N. Risgaard-Petersen, and S. Rysgaard. 1998. Interpretation of measured concentration profiles in sediment pore water. Limnol. Oceanogr. 43: 1500-1510.

Berg, P., H. Roy, and P. L. Wiberg. 2007. Eddy correlation flux measurements: The sediment surface area that contributes to the flux. Limnol. Oceanogr. 52: 1672-1684.

Berg, P., and M. Huettel. 2008. Monitoring the seafloor using the noninvasive eddy correlation technique: Integrated benthic exchange dynamics. Oceanography 21: 164- 167.

Berg, P., M. H. Long, M. Huettel, J. E. Rheuban, K. J. McGlathery, R. W. Howarth, K. H. Foreman, A. E. Giblin, and R. Marino. 2013. Eddy correlation measurements of oxygen fluxes in permeable sediments exposed to varying current flow and light. Limnol. Oceanogr. 58: 1329-1343.

Blicher, M. E., M. K. Sejr, and S. Rysgaard. 2009. High carbon demand of dominant macrozoobenthic species indicates their central role in ecosystem carbon flow in a sub- Arctic fjord. Mar. Ecol. Prog. Ser. 383: 127-140.

Boetius, A., and others. 2013. Export of algal biomass from the melting Arctic sea ice. Science 339: 1430-1432.

Boudreau, B. P. 2001. Solute transport above the sediment-water interface. In Boudreau B. P. and B. B. Jørgensen [eds.], The Benthic Boundary Layer. Oxford University Press. pp. 104-123.

Brand, A., D. F. McGinnis, B. Wehrli, and A. Wüest. 2008. Intermittent oxygen flux from the interior into the bottom boundary of lakes as observed by eddy correlation. Limnol. Oceanogr. 53: 1997-2006.

Burba, G. 2013. Eddy covariance method for scientific, industrial, agricultural, and regulatory applications: A field book on measuring ecosystem gas exchange and areal emission rates. LI-COR Biosciences.

Cahoon, L. B. 1999. The role of benthic microalgae in neritic ecosystems. Oceanogr. Mar. Biol. Annu. Rev. 37: 47–86.

References

Canfield, D. E., 1989. Sulfate reduction and oxic respiration in marine sediments: implications for organic carbon preservation in euxinic environments. Deep-Sea Research 36: 121- 138.

Canfield, D. E., B. B. Jørgensen, H. Fossing, R. N. Glud, J. K. Gundersen, N. B. Ramsing, et al. 1993. Pathways of organic carbon oxidation in three continental margin sediments. Marine Geology 113: 2740. Canfield, D. E., E. Kristensen, and B. Thamdrup. 2005. Aquatic Geomicrobiology. Elsevier. Charette, M. A., and W. H. F. Smith. 2010. The volume of Earth’s oceans. Oceanography 23: 112-114. Chipman, L., D. Podgorski, S. Green, J. Kostka, W. Cooper, and M. Huettel. 2010. Decomposition of plankton-derived dissolved organic matter in permeable coastal sediments. Limnol. Oceanogr. 55: 857-871.

Chipman, L., M. Huettel, P. Berg, V. Meyer, I. Klimant, R. N. Glud, and F. Wenzhoefer. 2012. Oxygen optodes as fast sensors for eddy correlation measurements in aquatic systems. Limnol. Oceanogr.: Methods 10: 304–316.

Cook, P. L. M., and H. Røy. 2006. Advective relief of CO2 limitation in microphytobenthos in highly productive sandy sediments. Limnol. Oceanogr. 51:1594-1601.

Cook, P. L. M., F. Wenzhöfer, R. N. Glud, F. Janssen, and M. Huettel. 2007. Benthic solute exchange and carbon mineralization in two shallow subtidal sandy sediments: Effect of advective pore-water exchange. Limnol. Oceanogr. 52: 1943-1963.

Dade, B. W., A. J. Hogg, and B. P. Boudreau. 2001. Physics of flow above the sediment-water interface. In Boudreau B. P. and B. B. Jørgensen [eds.], The Benthic Boundary Layer. Oxford University Press.

Dalsgaard, T. 2003. Benthic primary production and nutrient cycling in sediments with benthic microalgae and transient accumulation of macroalgae. Limnol. Oceanogr. 48: 2138-2150.

Dame, R. F., 1996. Ecology of marine bivalves: an ecosystem approach. CRC Press.

Davoult, D., and F. Gounin. 1995. Suspension-feeding activity of a dense Ophiothrix fragilis (Abildgaard) population at the water-sediment interface: time coupling of food availability and feedinb behavior of the species. Estuarine, Coastal and Shelf Science 41: 567-577.

Duarte, C. M., J. J. Middelburg, and N. Caraco. 2005. Major role of marine vegetation on the oceanic carbon cycle. Biogeosciences 2: 1-8.

References

Emerson, S., and J. Hedges. 2003. Sediment diagenesis and benthic flux. Treatise on Geochemistry 6: 293-319.

Enríquez, S., and A. Rodríguez-Román. 2006. Effect of water flow on the photosynthesis of three marine macrophytes from a fringing-reef lagoon. Mar. Ecol. Prog. Ser. 323: 119- 132.

Epping, E. H. G., and B. B. Jørgensen. 1996. Light-enhanced oxygen respiration in benthic phototrophic communities. Mar. Ecol. Prog. Ser. 139: 193–203.

Falkowski, P. G., and J. A. Raven. 2006. Aquatic photosynthesis. Princeton University Press.

Fenchel, T., and B. B. Jørgensen. 1977. Detritus food chains of aquatic systems: the role of bacteria. In: Alexander M, editor. Advances in Microbial Ecology 1. Plenum Press. p 1- 58.

Fenchel, T., and R. N. Glud. 2000. Benthic primary production and O2 - CO2 dynamics in a shallow-water sediment: Spatial and temporal heterogeneity. Ophelia 53: 159–171.

Férron, S. and others. 2008. Hydrodynamic characterization and performance of an autonomous benthic chamber for use in coastal systems. Limnol. Oceanogr. Methods 6: 558-571.

Gattuso, J. P., B. Gentili, C. M. Duarte, J. A. Kleypas, J. J. Middelburg, and D. Antoinne. 2006. Light availability in the coastal ocean: Impact on the distribution of benthic photosynthetic organisms and their contribution to primary production. Biogeosciences 3: 489–513.

Glud, R. N. and N. Blackburn. 2002. The effect of chamber size on benthic oxygen uptake measurements: a simulation study. Ophelia 56: 23-31.

Glud, R. N., J. K. Gundersen, H. Røy, and B. B. Jørgensen. 2003. Seasonal dynamics of benthic

O2 uptake in a semi-enclosed bay: importance of diffusion and fauna activity. Limnol. Oceanogr. 48: 1265-1276.

Glud, R. N., P. Berg, H. Fossing, and B. B. Jørgensen. 2007. Effect of the diffusive boundary

layer (DBL) on the benthic mineralization and O2 distribution: A theoretical modelling exercise. Limnol. Oceanogr. 52: 547–557.

Glud, R. N. 2008. Oxygen dynamics of marine sediments. Mar. Biol. Res. 4: 243–289.

Glud, R. N., J. Woelfel, U. Karsten, M. Kuhl, and S. Rysgaard. 2009. Benthic microalgal production in the Arctic: Applied methods and status of the current database. Bot. Mar. 52: 559-678.

References

Glud, R. N., P. Berg, A. Hume, P. Batty, M. E. Blicher, K. Lennert, and S. Rysgaard. 2010. Benthic oxygen exchange across hard bottom substrates quantified by eddy correlation in a sub-Arctic fjord. Mar. Ecol. Prog. Ser. 417: 1–12.

Glud, R. N., S. Rysgaard, G. Turner, D. F. McGinnis, and R. J. G. Leakey. 2014. Biological- and physical-induced oxygen dynamics in melting sea ice of the Fram Strait. Limnol. Oceanogr. 59: 1097-1111.

Hancke, K., B. K. Sorrell, L. C. Lund-Hansen, M. Larsen, T. Hancke, and R. N. Glud. 2014. Effects of temperature and irradiance on a benthic microalgal community: A combined two-dimensional oxygen and fluorescence imaging approach. Limnol. Oceanogr. 59: 1599-1611.

Huettel, M., and G. Gust. 1992. Impact of bioroughness on interfacial solute exchange in permeable sediments. Mar. Ecol. Prog. Ser. 89: 253–267.

Hume, A. C., P. Berg, and K. J. McGlathery. 2011. Dissolved oxygen fluxes and ecosystem metabolism in an eelgrass (Zostera marina) meadow measured with the eddy correlation technique. Limnol. Oceanog. 56: 86-96.

Holtappels M., R. N. Glud, D. Donis, B. Liu, A. Hume, F. Wenzhoefer, and M. Kuypers. 2013. Effect of transient bottom water currents and oxygen concentrations on benthic exchange rates as assessed by eddy correlation measurements. J. Geophys. Res.: Oceans 118: 1157-1169.

Hulthe, G., S. Hulth, and P. O. J. Hall. 1998. Effect of oxygen on degradation rate of refractory and labile organic matter in continental margin sediments. Geochimica et Cosmochimica Acta 62:1319-1328. IOCCG. 2000. Remote Sensing of Ocean Colour in Coastal, and Other Optically-Complex Waters. Sathyendranath, S. (ed.), Reports of the International Ocean-Colour Coordinating Group, No. 3, IOCCG, Dartmouth, Canada.

Jahnke, R. A., J. R. Nelson, M. E. Richards, C. Y. Robinson, A. M. F. Rao, and D. B. Jahnke. 2008. Benthic primary productivity on the Georgia midcontinental shelf: Benthic flux measurements and high-resolution, continuous in situ PAR records. Journal of Geophysical Research 113: C08022. doi:10.1029/2008JC004745.

Jassby, A. D., and T. Platt. 1976. Mathematical formulation of relationship between photosynthesis and light for phytoplankton. Limnol. Oceanogr. 21: 540–547

Jørgensen, B. B., R. N. Glud, and O. Holby. 2005. Oxygen distribution and bioirrigation in Arctic fjord sediments (Svalbard, Barents Sea). Marine Ecology Progress Series 292: 85- 95.

References

Kühl, M., R. N. Glud, H. Ploug, and N. B. Ramsing. 1996. Microenvironmental control of photosynthesis and photosynthesis-coupled respiration in an epilithic cyanobacterial biofilm. J. Phycol. 32: 799–812.

Kristensen, E., G. Penha-Lopes, M. Delefosse, T. Valdemarsen, C. O. Quintana, and G. T. Banta. 2012. What is bioturbation? The need for a precise definition for fauna in aquatic sciences. Marine Ecology Progress Series 446: 285-302.

Kristensen, E., and M. Holmer. 2001. Decomposition of plant materials in marine sediments - 2- exposed to different electron acceptors O2, NO3 , SO4 , with emphasis on substrate origin, degradation kinetics, and the role of bioturbation. Geochimica et Cosmochimica Acta 65:419-433. Kristensen, E., and J. E. Kostka. 2005. Macrofaunal burrows and irrigation in marine sediment: microbiological and biogeochemical interactions. In S. Fagherazzi, M. Marani, and L. K. Blum [eds.], The Ecogeomorphology of Tidal Marshes. American Geophysical Union.

Lee, X., W. Massmann, and B. Law. 2004. Handbook of micrometeorology: A guide for surface flux measurement and analysis. Kluwer Academic Publishers.

Long, M. H., D. Koopmans, P. Berg, S. Rysgaard, R. N. Glud, and D. H. Søgaard. 2012. Oxygen exchange and ice melt measured at the ice-water interface by eddy correlation. Biogeosciences 9: 1957-1967.

Long, M. H., P. Berg, D. de Beer, and J. C. Zieman. 2013. In situ coral reef oxygen metabolism: An eddy correlation study. PloS one 8: e58581, doi:10.1371/journal.pone.0058581

Lorke, A., B. Müller, M. Maerki, and A. Wüest. 2003. Breathing sediments: the control of the diffusive transport across the sediment-water interface by periodic boundary-layer turbulence. Limnol. Oceanogr. 48: 2077-2085.

Lorrai, C., D. F. McGinnis, P. Berg, A. Brand and A. Wüest. 2010. Application of oxygen eddy correlation in aquatic systems. J. Atmosph. Ocean. Tech. 27: 1533-1546.

Markager, S., and K. Sand-Jensen. 1992. Light requirements and depth zonation of marine macroalgae. Marine Ecology Progress Series 88: 83-92.

Martin, S., J. Clavier, L. Chauvaud, and G. Thouzeau. 2007. Community metabolism in temperate maerl beds. I. Carbon and carbonate fluxes. Mar. Ecol. Prog. Ser. 335: 19- 29.

Mass, T., A. Genin, U. Shavit, M. Grinstein, and D. Tchernov. 2010. Flow enhances photosynthesis in marine benthic autotrophs by increasing the efflux of oxygen from the organism to the water. Proc. Nat. Ac. Sci. 107: 2527-2531.

References

McGee, D., R. A. Laws, and L. B. Cahoon. 2008. Live benthic diatoms from the upper continental slope: Extending the limits of marine primary production. Mar. Ecol. Prog. Ser. 356: 103-112.

McGinnis, D. F., S. Cherednichenko, S .Sommer, P. Berg, L. Rovelli, R. Schwarz, R. N. Glud, and P. Linke. 2011. Simple, robust eddy correlation amplifier for aquatic dissolved oxygen and hydrogen sulfide flux measurements. Limnol. Oceanogr.: Methods 9: 340-347.

McGinnis, D. F., S. Sommer, A. Lorke, R. N. Glud, and P. Linke. 2014. Quantifying tidally- driven benthic oxygen exchange across permeable sediments: An aquatic eddy correlation study. Journal of Geophysical Research. doi: 10.1002/2014JC010303.

Middelburg, J. J., and K. Soetaert. 2004. The role of sediments in shelf ecosystem dynamics, p. 353–375. In K. H. B. Allan and R. Robinson [eds.], The global coastal ocean. Harvard University Press.

Reidenbach, M. A., P. Berg, A. Hume, J. C. R. Hansen, and E. R. Whitman. 2013. Hydrodynamics of intertidal oyster reefs: the influence of boundary layer flow processes on sediment and oxygen exchange. Limnol. Oceanog.: Fluids and Environments 3: 225-239.

Reimers, C. E., H. T. Ozkan-Haller, P. Berg, A. Devol, K. McCann-Grosvenor, and R. D. Sanders. 2012. Benthic oxygen consumption rates during hypoxic conditions on the Oregon shelf: Evaluation of the eddy correlation method. J. Geophys. Res. 117: C02021. doi: 10.1029/2011JC007564.

Revsbech N. P. 1989. An oxygen microelectrode with a guard cathode. Limnol. Oceanogr. 34: 474-478.

Rheuban, J. E., P. Berg, and K. J. McGlathery. 2014. Ecosystem metabolism along a colonization gradient of eelgrass (Zostera marina) measured by eddy correlation. Limnol. Oceanogr. 59: 1376-1387.

Sun, M-Y., R. C. Aller, C. Lee, and S. G. Wakeham. 1999. Enhanced degradation of algal lipids by benthic macrofaunal activity: Effect of Yolida limatula. Journal of Marine Research 57: 775-804.

Therkildsen, M. S., and B. Lomstein. 1993. Seasonal variation in net benthic C-mineralization in a shallow estuary. FEMS Microbiology Ecology 12:131-142.

Thomson, M. L., and L. C. Schaffner. 2001. Population biology and secondary production of the suspension feeding polychaete Chaetopterus cf. variopedatus: Implications for benthic-pelagic coupling in lower Chesapeake Bay. Limnol. Oceanogr. 46: 1899-1907.

References

Wenzhöfer, F., and R. N. Glud. 2004. Small-scale spatial and temporal variability in coastal

benthic O2 dynamics: Effects of fauna activity. Limnol. Oceanogr. 49: 1471-1481.

Wollast, R. 1991. The coastal organic carbon cycle: fluxes, sources and sinks, p. 365-381. In J.-M. M. R.-F.-C. Mantoura, R. Wollast [ed.], Ocean Margin Processes in Global Change. John Wiley & Sons.

Wüest, A., and A. Lorke. 2003. Small-scale hydrodynamics in lakes. Annu. Rev. Fluid Mech. 35: 373-412.

Appendix

2. APPENDIX

The work presented in this thesis includes only those studies that have been published, submitted, or are close to submission. Much work still remains to be done on other datasets that I collected during my PhD.

The aim of this section is to give a brief overview of these remaining datasets.

2.1.: Sea ice O2 dynamics in the Central Arctic during the 2012 sea ice minimum

K. Attard, F. Wenzhöfer, D. F. McGinnis, H. Sørensen, and R. N. Glud.

I joined a research expedition to the Central Arctic aboard R.V. Polarstern during the summer of 2012. The cruise visited the ice-covered eastern central basins between 82⁰ and 89⁰ N and 30⁰ to 130⁰ E. I conducted under-ice measurements of O2 exchange using EC and monitored the hydrodynamics in the under-ice boundary layer using a high-resolution ADCP during eight ‘ice stations’, each lasting 48 hrs (Fig. A1).

Fig. A1: Top left: The RV Polarstern docked on an ice floe (Photo by Stefan Hendricks). Bottom left: An EC O2 system deployed for under-ice measurements. Right: The 8 ice stations visited during the expedition (Boetius et al. 2013).

Appendix

-2 -1 Typically, the EC O2 exchange rates resolved a net O2 uptake by the ice of around 4-8 mmol O2 m d . In some cases, a light response was evident in the O2 data, although the extent to which this was driven by light or by physical processes such as flow velocity and ice melt remain to be verified through a more detailed analyses of the data. Some data examples are presented below (Fig. A2).

Fig. A2: A few preliminary data examples for under-ice EC O2 exchange rates measured during summer 2012 in the Central Arctic. The dataset in the top left panel is from Ice Station 5, the top right panel is from Ice Station 6, and the dataset in the bottom panel is from Ice Station 8.

Appendix

2.2.: River metabolism in the River Avon catchment, England

L. Rovelli, K. Attard, H. Stahl, M. Trimmer, and R. N. Glud.

We conducted a seasonal study in six rivers within the River Avon catchment representing a gradient of sediment types ranging from permeable carbonates to cohesive clays. The overall aim was to quantify the benthic metabolism in the rivers. This included EC measurements to estimate the benthic primary production and carbon turnover (Fig. A3), in situ water incubations to quantify the water column O2 production and consumption, meteorological data as well as water flow and discharge measurements.

Fig. A3: Deploying an EC instrument in the investigated rivers (top panels). Bottom panel: a typical EC time series and environmental parameters for one of the chalk rivers in spring.

Appendix

2.3.: Benthic community primary production and carbon turnover along a depth gradient in a high-Arctic fjord (Young Sound, northeast Greenland).

K. Hancke, K. Attard, M. Sejr, M. Blicher, S. Rysgaard, and R. N. Glud.

This expedition took place in July/August 2014 in Young Sound, northeast Greenland. The main aim was to quantify benthic community net O2 production and consumption as a function of irradiance and depth. EC deployments were carried out at 9 stations along a depth transect ranging from 5 to 160 m. The duration of the EC deployments ranged between 24-48 h at each station. High-resolution images of the seabed were taken to estimate the biomass of the dominant macrofauna (Fig. A4).

Fig. A4: Representative images from each station investigated in Young Sound, northeast Greenland.

Paper I Benthic primary production seasonality

4. PUBLICATIONS

4.1. Paper I

Seasonal rates of benthic primary production in a Greenland fjord

measured by aquatic eddy correlation

K. M. Attard, R. N. Glud, D. F. McGinnis, and S. Rysgaard

Limnology and Oceanography Vol. 59, 2014, 1555-1569

Paper I Benthic primary production seasonality

Seasonal rates of benthic primary production in a Greenland fjord measured by aquatic eddy-

correlation

Karl M. Attard,1,2,* Ronnie N. Glud,1,2,3,4 Daniel F. McGinnis,1,2,5,6 Søren Rysgaard2,4,7

1 University of Southern Denmark, Nordic Centre for Earth Evolution (NordCEE), Odense, Denmark

2 Greenland Institute of Natural Resources, Greenland Climate Research Centre, Nuuk, Greenland

3 Scottish Association for Marine Sciences, Scottish Marine Institute, Oban, United Kingdom

4 University of Århus, Arctic Research Centre, Århus, Denmark

5 Helmholtz Centre for Ocean Research Kiel (GEOMAR), Kiel, Germany

6 Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB), Berlin, Germany

7 University of Manitoba, Centre for Earth Observation Science, Clayton Riddell Faculty of

Environment, Earth, and Resources, Winnipeg, Manitoba, Canada

*Corresponding author: [email protected]

Running title: Benthic primary production seasonality

Paper I Benthic primary production seasonality

Acknowledgements

We thank Anni Glud, Heidi L. Sørensen, Lorenz Meire, Lars Heilmann, Flemming Heinrich, Thomas

Krogh, Martin E. Blicher, Thomas-Juul Pedersen, Ivali Lennert, Kunuk Lennert, and Knud

Kreutzmann for technical and/or field assistance, as well as Lorenzo Rovelli and Kasper Hancke for discussions on topics related to this manuscript. We thank Cynthia Bluteau and Andreas Lorke for providing routines for calculating the inertial dissipation and planar rotation, respectively. We thank Peter Berg and a second anonymous reviewer for constructive comments that improved this manuscript. This project was financed by the Commission for Scientific Research in Greenland

(KVUG; GCRC6507), the UK Natural Environmental Research Council (NERC, NE/F018612/1,

NE/F0122991/1, NE/G006415/1), the European Research Council through an Advanced Grant

(ERC-2010-AdG20100224), the Canada Excellence Research Chair program, and the Danish

National Research Foundation (DNRF53).

Paper I Benthic primary production seasonality

Abstract

We present the first year-round estimates of benthic primary production at four contrasting shallow (3-22 m depth) benthic habitats in a southwest Greenland fjord. In situ measurements were performed using the non-invasive aquatic eddy correlation (EC) oxygen (O2) flux method. A series of high-quality multiple-day EC datasets document the presence of a year-round productive benthic phototrophic community. The shallow water sites were on average autotrophic during the

-2 -1 spring and summer months, up to 43.6 mmol O2 m d , and heterotrophic or close to metabolic balance during the autumn and winter. Substantial benthic gross primary production (GPP) was measured year round. The highest GPP rates were measured during the spring, up to 5.7 mmol O2

-2 -1 -2 -1 -2 -1 m h (136.8 mmol O2 m d ), and even at low light levels (<80 μmol quanta m s ) during late

-2 -1 -2 -1 autumn and winter we measured rates of up to 1.8 mmol O2 m h (43.2 mmol O2 m d ) during peak irradiance. The benthic phototrophic communities responded seasonally to ambient light levels and exhibited year-round high photosynthetic efficiency. In situ downwelling irradiances as

-2 -1 low as ~2 μmol quanta m s induced an autotrophic response and light saturation indices ( ) were as low as 11 μmol quanta m-2 s-1 in the winter. On an annual timescale, the average areal

-2 -1 rate of benthic GPP was 11.5 mol O2 m yr which is ~1.4 times higher than the integrated gross pelagic primary production of the ~30-50 m deep photic zone of the fjord. These results document the importance of benthic photosynthesis on an ecosystem level and indicate that the benthic phototrophic compartment should be accounted for when assessing carbon and nutrient budgets as well as responses of coastal Arctic ecosystems to climate change.

Paper I Benthic primary production seasonality

Introduction

Shallow water environments constitute the most dynamic and productive ecosystems of the world’s oceans (Middelburg and Soetaert 2004; Glud 2008). The productivity in shallow ecosystems is driven by both pelagic and benthic primary production. Despite this pattern, most studies focus on the pelagic environment alone and potentially ignore a considerable contribution from the benthic communities (Cahoon 1999). Furthermore, the presence of oxygen (O2) producing benthic microalgae alters the availability of biologically-important solutes such as

- + nitrate (NO3 ), ammonium (NH4 ), O2 and carbon dioxide (CO2) within the surface sediments as well as in the overlying water column (Fenchel and Glud 2000; Dalsgaard 2003). Benthic algae are in a prime location to benefit from the nutrients that are released from the sediment following mineralization of organic matter, and provided sufficient light is available they compete successfully with phytoplankton for the regenerated nutrients. In oligotrophic coastal waters benthic primary production therefore tends to dominate ecosystem production (Jahnke et al.

2000; Glud et al. 2002; Middelburg and Soetaert 2004).

The Arctic and subarctic regions are sparsely populated and only locally affected by eutrophication, but the region is undergoing dramatic transformations. Changes in sea ice extent and thickness, precipitation and river discharge, and wind patterns alter the light and nutrient availability in the coastal zone and are expected to affect coastal productivity (Wassmann et al.

2011). While considerable research has focused on pelagic productivity in the Arctic, benthic primary production is surprisingly understudied (Glud et al. 2009). The extensive shelves of the

Arctic Ocean encompass some 25% of the global shelf areas and may host a substantial biomass of benthic primary producers (Gattuso et al. 2006).

Paper I Benthic primary production seasonality

Rates of in situ benthic primary production are most often inferred from measurements of O2 exchange between the seabed and the overlying water column (termed the ‘benthic O2 exchange rate’, Glud 2008). Traditionally, the benthic O2 exchange rate has been quantified using benthic chambers and microprofilers. Transparent benthic O2 chambers enclose a known sediment area

2 (typically ~0.2 m ) and the areal exchange rate of O2 is inferred from the change in O2 concentration over time. Benthic O2 microprofilers provide point measurements of the O2 concentration gradient at the sediment-water interface to estimate the ‘diffusive benthic O2 exchange’. Both methods are difficult to apply to hard benthic strata, which are widespread in

Arctic and subarctic systems, and as a result large regions of the coasts remain severely understudied (Glud et al. 2010).

The introduction of the ‘aquatic eddy-correlation’ (EC) technique for measuring benthic O2 fluxes by Berg et al. (2003) has made it possible to investigate the benthic O2 exchange rate under ‘true’ in situ conditions. The EC technique infers the O2 exchange at the sediment-water interface from continuous Eulerian, high-frequency measurements of the flow velocity and O2 concentration taken at the same point within the benthic boundary layer (BBL) close to the sediment surface

(Berg et al. 2003). This information is used to derive a continuous time-series of vertical O2 fluxes across a horizontal plane. Since the EC measurements are carried out away from the sediment surface, the EC method is not confined to soft sediments. Importantly, EC fluxes integrate a large area of the seabed (typically 10-100 m2) under the natural light and flow conditions (Berg et al.

2007). To date, coastal applications of the EC technique constitute a limited but growing database.

EC measurements on hard-bottom benthic surfaces (Glud et al. 2010), seagrass beds (Hume et al.

2011), permeable sediments (Berg et al. 2013), sea ice (Long et al. 2012), coral reefs (Long et al.

Paper I Benthic primary production seasonality

2013), and oyster beds (Reidenbach et al. 2013) document the validity of this method in environments where benthic chamber and microprofile measurements would be compromised.

The present study draws on the advantages of the EC method to estimate rates of benthic primary production and carbon turnover in shallow contrasting benthic ecosystems in a south west

Greenland fjord. We monitored these sites over a 13 month period to assess seasonal trends in the rates of benthic O2 exchange and inferred rates of benthic productivity from a series of multiple-day datasets. The objective of this study was to consider diel, day-to-day, as well as seasonal changes in benthic productivity, and to assess the ecological importance of benthic primary production in a vastly understudied region that is undergoing rapid climatic changes.

Methods

Study sites

The study was conducted between May 2011 and June 2012 within the Godthåbsfjord system in south-west Greenland (Fig. 1). A long-term monitoring program has been in place in Godthåbsfjord since 2006 with the aim to establish a long time-series for key physical, chemical, and biological oceanographic observations (MarineBasis, www.nuuk-basic.dk).

The EC system was deployed on a regular basis throughout the year at three locations nearby

Nuuk, Greenland’s capital city. Furthermore, we present a single deployment from a deeper fourth site. In all cases the EC system was lowered by hand from a small research vessel. Representative images from each of the four sites are shown in Fig. 2.

Site 1 was located in Nipisat Sound, a protected inlet with an area of around 3 km2. The EC system was placed in the inner part of the inlet at a depth of 3 m at mean low water. The flow in Nipisat

Paper I Benthic primary production seasonality

Sound was tidally-driven, with typical flow velocities ranging from 2 to 10 cm s-1 over a tidal cycle.

The sediments in the sound were spatially heterogeneous ranging from silty sands on the flanks to winnowed sands in the middle of the sound. Still and video imaging along with sediment sampling documented the presence of year-round benthic microalgae, while patches of filamentous green algae and small vascular brown macroalgae extending 2-10 cm up into the water column were present during the summer months. The sediment hosted dense infauna and epifauna communities dominated by polychaetes, bivalves and echinoderms (Blicher et al. 2011). The inner regions of the sound were seasonally ice covered during the winter with a thickness of ~ 10 cm.

Sites 2, 3, and 4 were located within Kobbefjord, a well-confined 25 km2 fjord close to the capital

Nuuk. Site 2 consisted of a sand flat with consolidated sands and gravels that extended for several hundred square meters within a depth range of 1-10 meters. Benthic primary producers consisted mainly of benthic microalgae and filamentous green and small vascular brown macroalgae. The water velocity typically ranged from 1-6 cm s-1 over a tidal cycle.

Site 3 consisted of a rocky embayment ranging in depth from 1-7 m. The interstices between the rocks were filled with coarse gravels and relict shell fragments. Corraline red algae covered many of the rocks, and high densities of the urchin Strongylocentrotus droebachiensis were observed on the rocks as well as on the gravel interstices. The water velocity typically ranged from 1-4 cm s-1 over a tidal cycle.

Site 4 was located close to the head of Kobbefjord at a depth of 22 m. The site was characterized by muddy sediments with occasional glacial drop stones encrusted in corraline algae. At the time of investigation (October 2011) large densities of the urchin S. droebachiensis were observed

Paper I Benthic primary production seasonality

actively grazing on the microphytobenthic mat as well as on the corraline algae. The water velocity at this site ranged from 1-4 cm s-1.

Eddy-correlation measurements

The setup of the applied EC system was similar to the original design by Berg et al. (2003). The main components of the system consisted of an Acoustic Doppler Velocimeter (ADV, Vector,

Nortek A.S.) and two parallel Clark-type O2 microsensors (10-20 µm tip diameter with a 90% response time ≤ 0.3 s; Revsbech 1989) that relay the signal to the ADV via two custom build submersible amplifiers (McGinnis et al. 2011). The deployment of two parallel O2 microsensors greatly increased our deployment success rate and provided an internal quality check of the O2 data (McGinnis et al. 2011).

The ADV recorded the streamwise (u), traverse (v), and vertical (w) flow velocity components as well as the O2 microsensor data at frequencies of 32 or 64 Hz and in addition collected ancillary information such as instrument pitch and roll, flow direction, and signal strength. The equipment was mounted onto a stainless-steel tripod frame measuring 130 cm by 90 cm, designed to minimize hydrodynamic interference. The ADV was mounted downward-facing perpendicular to the seabed surface and the ADV sampling volume was located 25 cm above the sediment surface.

This distance was at least 2 times larger than the stones, benthic epifauna and other bottom features that we observed. The microsensor tips were carefully positioned at the edge of the ADV sampling volume to extract the O2 data close to the ADV measurement point without compromising the velocity measurements.

Our EC system was also equipped with a conductivity-temperature-depth (CTD) sensor (SBE 19 plus V2, Seabird). Apart from the standard suite of sensors, the CTD had a downwelling

Paper I Benthic primary production seasonality

photosynthetically active radiation (PAR) sensor (QCP-2000, Biospherical Instruments) and an O2 optode (4340, Aanderaa). This instrument provided continuous measurements of the environmental parameters covering each EC deployment at a resolution of 60 or 120 seconds.

Prior to deployment, the EC O2 microsensors were left to polarize for 10-12 hours to minimize sensor drift during deployments. A preliminary calibration of the sensors was carried out using two water samples of known O2 concentration at in situ temperature and salinity. A sodium dithionite solution was used for the zero O2 saturation value, and collected bottom water samples at the measuring site were used for the in situ saturation value. The O2 concentration was later determined in the laboratory by Winkler titration. In total 18 successful deployments were made during our 13 month measurement campaign, with the deployment time ranging from 20 to 119 hours with a total of ~750 hours and an overall deployment average of 41 hours. Unfavorable sea ice conditions persisted during spring 2012 and prevented access to the sites in Kobbefjord, limiting our observations during this period.

Eddy-correlation fluxes

The EC O2 fluxes were first extracted from the raw EC dataset and then evaluated for their ‘quality’ based on a set of defined criteria. Flux extraction was carried out using the software package

Sulfide Oxygen Heat Flux Eddy Analysis (SOHFEA) version 2.0 (available at www.dfmcginnis.com/SOHFEA). SOHFEA employs a protocol for flux extraction that is very similar to the one described by Lorrai et al. (2010). Additional data treatments that are detailed below but not available in SOHFEA were carried out in MATLAB® (MathWorks). The datasets were processed for flux extraction in the following order: 1) Weak signals in the raw 32 or 64 Hz ADV velocity data were identified by a low beam correlation and/or low signal strength. Individual data points with

Paper I Benthic primary production seasonality

beam correlations below 70% and signal-to-noise ratios below 12 dB were discarded. These thresholds were lowered to 50% and 10 dB respectively during the winter months, when the particle density in the water was greatly reduced. 2) The 32 or 64 Hz raw EC data was averaged down to 8 Hz. This reduced the noise level and the smaller file size allowed for easier data handling. 3) A spectral analysis was carried out on the 8 Hz vertical velocity ( ) and the O2 microsensor data ( ). The spectra showed the presence of an inertial subrange, identified as the region on the pressure spectral density plot for and where the slope of energy cascade to the smaller scales followed the predicted -5/3 fit, suggesting well-developed turbulence. The spectra furthermore revealed that the data reduction through adjacent averaging from 64 or 32 Hz to 8 Hz did not result in a loss of signal at high frequency, since most of the turbulent contributions typically occurred at a frequency of 2 Hz or lower. 4) The O2 microsensors were calibrated to the

CTD O2 optode data. 5) Spike noise in the velocity and O2 microsensor data were removed using the 3D phase space method by Mori et al. (2007). 6) The measured 8 Hz ADV velocity dataset was rotated using the ‘planar rotation’ method to obtain a vertical velocity component that is normal to the local streamline (Lorke et al. 2013). Rotation of the dataset is required to exclude vertical projections of the horizontal velocity components that may arise from a slight tilt in the

-2 -1 instrument or from topographic features. 7) The O2 fluxes, in mmol m h , were extracted from the ADV velocity and O2 microsensor data as the covariance 〈 〉, where and are deviations from a least-squares linear trend fitted to the measured vertical velocity and O2 concentration respectively, and the angle bracket denotes time averaging. Following Reynolds decomposition theory the vertical velocity vector may therefore be expressed as 〈 〉 and the scalar quantity 〈 〉 (Berg et al. 2003). The selected time averaging interval is a tradeoff between including as many of the flux-contributing turbulent eddies as possible while excluding

Paper I Benthic primary production seasonality

low-frequency non-turbulent contributions such as advective flows that may compromise covariance statistics (McGinnis et al. 2008). To determine the optimal time interval, an analysis was carried out to investigate the effects of averaging time on the mean covariance and subsequent flux estimates for the friction velocity ( ) and the O2 fluxes. The is directly related to the turbulence regime within the BBL. Estimates for were computed from complex Reynolds stress measurements derived from the ADV velocity time series (McPhee 2008). The streamwise

(u), traverse (v), and vertical (w) velocity components were decomposed into a mean and deviatory velocity as 〈 〉 , 〈 〉 and 〈 〉 . The was then calculated

as 〈 〉 〈 〉 . The mean and O2 flux were computed as a function of the ensemble average for periods of both low and high flow velocity magnitude. A time window of 10 min was consistently identified as the optimal interval for flux calculation at the four sites (Fig. 3).

8) A time-shift correction was applied to the data. Time shifting was performed for each ensemble interval by shifting the O2 data in time relative to the velocity data to a maximum of 2 seconds to achieve the maximum correlation (defined in terms of the maximum flux) for 〈 〉. This correction is necessary when the physical separation between the O2 sensor and the ADV measurement volume, and/or the sensor response time, result in a slight misalignment in the data

(McGinnis et al. 2008). 9) By assuming law-of-the-wall velocity profiles, the mean sediment surface

roughness ( ) was estimated as , where is the measurement height above

the benthic surface (0.25 m), ҡ is the von Karman constant (0.41), and is the flow velocity

magnitude. 10) The coefficient of drag (C ) was computed as (McPhee 2008). D

The extracted 10 min EC fluxes were then evaluated for their quality based on three criteria, namely: 1) collisions of particles or debris with the O2 microsensor (spikes), 2) rapid changes in

Paper I Benthic primary production seasonality

flow direction, and 3) the level of anisotropic turbulence, computed as the ratio between the horizontal and vertical dissipation rate of turbulent kinetic energy (ε). We estimated the ε using the inertial subrange method by Bluteau et al. (2011), which is independent of law-of-the-wall scaling. Typically, stratification build-up under periods of low flow velocity reduces vertical turbulent mixing to a greater extent relative to the horizontal counterpart. The O2 fluxes as measured by EC are seen to be highly suppressed, indicating that turbulence transport is also severely suppressed, and the eddy assumptions are no longer valid (Brand et al. 2008). The threshold anisotropy ratio used to filter EC O2 fluxes was typically 5-15, although it varied between sites and between individual deployments. Altogether the screening process based on the three criteria typically filtered out <20% of the measured 10 min EC O2 fluxes. An example of how factors like low flow velocity and sensor collisions may compromise EC O2 exchange rate calculations is illustrated in Fig. 4.

Benthic community light response

The light dependency of the EC O2 fluxes was evaluated using the photosynthesis vs. irradiance (P-

I) relationship (Jassby and Platt 1976). The screened 10 min EC O2 fluxes covering at least one diel period were averaged down to longer time intervals (1-4 h) to reduce the short-term variability that was independent of the PAR. The averaged fluxes were then plotted against the PAR data.

-2 -1 The relationship between the daytime net ecosystem production (NEP, in mmol O2 m h ) and the

-2 -1 irradiance (I, in μmol quanta m s ) was estimated as ( ) , where is the

-2 -1 maximum rate of gross primary production (GPP, in mmol O2 m h ), is the light saturation

-2 -1 -2 -1 parameter (in μmol quanta m s ), and R is the nighttime respiration term (in mmol O2 m h ).

The fitting parameters , , and R were allowed to vary until a least-squares line of best fit to the

Paper I Benthic primary production seasonality

data was obtained. This simple model adequately fit all the datasets and in most cases showed a remarkably tight fit to the measured EC data (see below). The compensation irradiance ( , in μmol quanta m-2 s-1) was subsequently derived from the P-I curves as the x-intercept.

Rates of benthic productivity

The rates of benthic productivity were calculated from each time series of the EC O2 exchange rates. Nighttime periods were identified as the periods when no response of the EC O2 exchange to the available PAR could be observed. The PAR threshold used to define nighttime periods was within the range of 0-10 µmol m-2 s-1. Daytime periods comprised the remaining intervals. Gaps in the data that occurred as a result of the flux screening process were filled with modeled fluxes based on the P-I relationship. The NEP and R were then derived from the gap-filled time series in

-2 -1 mmol O2 m h as a bulk average of the O2 fluxes during the daytime and nighttime, respectively.

-2 Assuming a light-independent respiration rate we also estimated the benthic GPP (in mmol O2 m hr-1) as | |.

The autotrophic-heterotrophic balance of the benthic ecosystem (termed ‘net ecosystem metabolism’, NEM, in mmol m-2 d-1) was derived as a weighted average of the NEP and R fluxes.

The NEM indicated whether sediment O2 production through photosynthesis balanced the various heterotrophic processes that directly or indirectly consume O2. Positive NEM values indicated a net O2 release by the benthic ecosystem (autotrophy) while negative NEM values indicated a net

O2 uptake (heterotrophy) over a 24 hour period.

Paper I Benthic primary production seasonality

Results

Metadata

This study is based on 18 EC deployments that in total integrate 750 hours of measuring time divided between the shallow water stations. During the study the salinity in the bottom waters ranged from 28.3 to 33.3 and the bottom water temperature ranged from 5.5°C during the summer to -1.4°C during the winter. Changes in temperature and salinity during the individual deployments were minimal. Integrated downwelling PAR measured in situ ranged from 14,105 mmol quanta m-2 d-1 in the summer to 103 mmol quanta m-2 d-1 in the winter. The bottom water was on average supersaturated with O2 during the summer (up to 127%) and undersaturated during the winter (down to 92%) (Table 1).

Eddy-correlation fluxes

The EC O2 flux time series typically exhibited a distinct diel structure in response to the availability of PAR (Fig. 5). Elevated PAR during the daytime typically resulted in positive EC O2 fluxes and a gradual increase in O2 concentration in the bottom water. Nighttime periods were characterized by negative EC O2 fluxes. The two main drivers of the EC-resolved O2 exchange rates were the light

(PAR) availability and the flow velocity at the seabed (Figs. 5, 6). The benthic communities exhibited high photosynthetic efficiency. The P-I relationship (Fig. 6, Table 2) demonstrated that little light was required to drive an autotrophic response and the benthic communities quickly became light saturated. Photoinhibition was not observed in any of the 18 datasets.

The diel signal in the data was overlain with short-term variability, seen in the 10 min EC exchange measurements in Figs. 5, 6. As an example, an increase in the flow from 2 to 12 cm s-1 during the

Paper I Benthic primary production seasonality

-2 -1 day enhances the EC O2 exchange by 3.5 times, from 2.1 to 7.4 mmol O2 m h . Similarly an

-1 increase in flow from 2 to 8 cm s during the night enhances the rate of O2 exchange from -0.1 to -

-2 -1 2.7 mmol O2 m h (Fig. 6). Although the short-term variability of the EC exchange rates with flow does not necessarily scale with metabolic processes in the sediment, rates that are integrated over longer time periods provide robust measures of the benthic O2 exchange rates and therefore the rate of carbon turnover- see Discussion.

Rates of benthic O2 exchange

We derived robust measurements for NEP, R, NEM, and GPP over the year from the high-quality

EC measurements at the four sites as evidence for an active and productive phototrophic benthic community (Fig. 7, Table 2). The highest rates of O2 exchange were observed during the spring and summer months. Sites 1 and 2 were autotrophic during these periods, while site 3 was heterotrophic. In all cases a substantial primary production was detected. Rates for benthic GPP were highly variable between the three sites during the spring and summer, ranging from 0.63 to

-2 -1 5.68 mmol O2 m h . Site 1 was overall the most productive of the three sites during the spring

-2 -1 and summer, with a combined 2011 and 2012 average rate of GPP of 2.99 ± 1.96 mmol O2 m h

-2 -1 (mean ± standard deviation (SD), n=5). For site 2 it was 1.48 ± 0.51 mmol O2 m h (mean ± SD,

-2 -1 n=3) and for site 3 it was 1.26 ± 0.49 mmol O2 m h (mean ± SD, n=3). The combined 2011 and

2012 average benthic GPP rate for the 3 sites during spring and summer was 2.11 ± 1.54 mmol O2 m-2 h-1 (mean ± SD, n=11).

-2 -1 High average rates of GPP up to 1.28 mmol O2 m h were maintained during the early autumn.

Under greatly reduced light availability during the late autumn and winter period (up to 100 times lower integrated PAR fluxes compared to the spring and summer months) the benthic primary

Paper I Benthic primary production seasonality

production was significantly reduced, but benthic phototrophs were still active. Light levels of 5

μmol quanta m-2 s-1 or less often resulted in an autotrophic response (Table 2, Fig. 8). The highest rates of GPP during this period were measured at site 2 in February 2012 (deployment 14), with an

-2 -1 average rate of 0.93 mmol O2 m h for the 50 hour deployment. However, during peak

-2 -1 irradiance, GPP rates were seen to be around 1.8 mmol O2 m h (Fig. 8). The late autumn and

-2 winter average rate for the three sites (deployments 11 through 14) was 0.65 ± 0.56 mmol O2 m h-1 (mean ± SD, n=4). Although all of our datasets in the late autumn and winter suggest autotrophic conditions during the daytime, the periods with adequate levels of PAR were restricted to only a few hours per day (Table 2, Figs. 7, 8). As a result, when integrated over 24 hours, the 3 sites were heterotrophic or at metabolic balance during the late autumn and winter

-2 -1 period (range from 0.8 to -2.7 mmol O2 m d , Table 2).

To demonstrate the benthic phototrophic potential of deeper waters a single 53 h deployment was carried out at 22 m water depth in October 2011 (deployment 9; Fig. 8). The average for this deployment was 4.3 μmol quanta m-2 s-1 which was the lowest of the 18 deployments. The rate of GPP was comparable to that observed at the shallow water stations during the winter (0.73

-2 -1 mmol O2 m h ).

Discussion

Benthic primary production

We observed a substantial benthic primary production in four contrasting benthic habitats in a

Greenland fjord over the year. Sites 1 and 2 were autotrophic during the spring and summer, with

-2 -1 NEM rates of up to 43.6 mmol O2 m d . Site 3 was heterotrophic year round, with NEM rates

-2 -1 increasing in magnitude with the transition from winter (-1.6 mmol O2 m d ) to summer (-12.3

Paper I Benthic primary production seasonality

-2 -1 mmol O2 m d ). Measurements by Glud et al. (2010) in nearby embayments in spring 2009 using

-2 -1 -2 -1 the EC method suggest an average GPP rate of 1.13 mmol O2 m h (27 mmol O2 m d ) which falls within the range of values obtained by our more comprehensive study (0.63 to 5.68 mmol O2 m-2 h-1). Although greatly reduced, benthic GPP during the late autumn and winter was detectable

-2 -1 in all cases, ranging from 0.05 to 1.28 mmol O2 m h . Benthic communities inhabiting deeper waters showed capabilities similar to the winter communities. Despite peak downwelling

-2 -1 irradiances of less than 20 μmol quanta m s , the EC O2 exchange rates demonstrate a well- acclimated phototrophic community capable of efficiently photosynthesizing at very low light levels.

Early estimates compiled by Cahoon (1999) suggest very modest rates of benthic primary production in the Arctic region (defined as 60-90° latitudes) with an average value of ~ 3.7 mmol m-2 d-1 for the 0-5 meter depth range. A more recent review by Glud et al. (2009) considered a broader database for the Arctic. From this the authors argue that the rates presented by Cahoon

(1999) most likely underestimate the true rates by 5-10 times. Our derived rate estimate for benthic GPP of 31.6 mmol m-2 d-1 is consistent with this observation (Table 2). This finding suggests that the difference in benthic GPP rates between the Arctic and lower latitudes as compiled by

Cahoon (1999) could be due to undersampling rather than true latitudinal related gradients.

Furthermore, given the potential for benthic primary production in the extensive shelf areas of the

Arctic (Gattuso et al. 2006), benthic primary production could become quantitatively more important for sustaining shallow-water food webs under future scenarios of reduced sea ice cover during summer (Glud et al. 2009).

Rates of benthic GPP that are derived from O2 exchange measurements assume equal O2 consumption during the day and night. However, O2 consumption during the daytime typically

Paper I Benthic primary production seasonality

exceeds nighttime O2 consumption by up to 1.8 times (Fenchel and Glud 2000) as a result of deeper O2 penetration during the daytime (hence a larger volume of sediment that can support aerobic mineralization), and an enhanced microbial turnover of leached labile photosynthesates

(Epping and Jørgensen 1996; Fenchel and Glud 2000). Therefore, rates of benthic GPP derived from O2 exchange rates most likely underestimate the true activity. Albeit conservative, the average areal rate of benthic GPP of 138 g C m-2 yr-1 is 1.4 times higher than the 2006-2011 mean depth-integrated gross pelagic production of the ~30-60 m deep photic zone of the region as

14 - -2 -1 assessed by in situ H CO3 incubations (96 g C m yr ; Rysgaard et al. 2012). On an ecosystem level, benthic photosynthesis therefore contributes significantly to primary production in the fjord and may serve as the main food source for shallow water macrofauna communities. Benthic fauna are an important pathway for the turnover of carbon and on average account for 25-50% of the total benthic O2 uptake in shallow waters (Glud 2008). The fauna mediated O2 uptake (FOU) is partitioned between the respiration of the organisms, which constitutes ~20-40% of the FOU, and the stimulated microbial mediated respiration and reoxidation following fauna irrigation and sediment reworking (Glud 2008). The urchin S. droebachiensis is the dominant macrobenthic species in waters shallower than 20 meters in the Kobbefjord region, and its carbon demand for depths of ~5 m at site 3 has been estimated at ~ 61 g C m-2 yr-1 (Blicher et al. 2009). Our estimates

-2 -1 -2 -1 for benthic primary production of 138 g C m yr (11.5 mol O2 m yr ) point towards the benthic phototrophic compartment as an important year-round nutritional source for the macrobenthic fauna in the fjord.

Paper I Benthic primary production seasonality

Drivers of eddy-correlation resolved O2 fluxes

Light availability

The availability of light at the sediment surface is the primary requirement for benthic photosynthesis, since the supply of nutrients is usually maintained by organic matter mineralization processes within the sediment (Barranguet et al. 1998). The derived P-I relationships (Table 2) clearly reflect a high photosynthetic efficiency of the benthic communities.

Even during winter at low light levels, the sediment exhibited a clear light response with values

-2 -1 -2 -1 as low as 2 μmol quanta m s and values as low as 11 μmol quanta m s . Photoautotrophic communities inhabiting light-limited systems such as high-latitude fjords and continental slopes exhibit very efficient light-harvesting capabilities that allow the communities to persist under greatly reduced irradiances (Kühl et al. 2001; McGee et al. 2008). From the P-I relations we estimate that the benthic communities at the three sites during the summer were light saturated for ~60% of the daytime, equivalent to ~40% of the diel period. Light saturation also persisted during winter, although the short days combined with the low angle of the sun restricted its duration to ~20% of the diel period.

Detailed studies on the photosynthetic performance of benthic systems indicate a dynamic community of microalgae that actively migrate within the sediment in response to the steep and varying ambient gradients of irradiance (Barranguet et al. 1998; Kromkamp et al. 1998). The dynamic behavior of benthic microalgae allows the community to maintain a high net productivity under prolonged periods of elevated irradiance. Consistent with our observations, net photoinhibition of benthic communities at high irradiance is not often observed in benthic systems.

Paper I Benthic primary production seasonality

Rates of benthic primary production showed consistent seasonal trends, with the derived P-I parameters , , and R changing with the availability of PAR. The benthic communities therefore exhibited a clear seasonal response in their photosynthetic capabilities. However, substantial variability in the P-I parameters between the sites within the same season and throughout the year is also evident (Fig. 9). There are obvious challenges when attempting to attribute changes in the photosynthetic performances of mixed benthic communities, often having species-specific responses (Kühl et al. 2001), to seasonal environmental variables such as temperature and PAR.

We noted a changing macroscopic composition of the benthic phototrophic communities at the shallow sites over the year. In the late spring and summer months both macroalgae and microalgae were present, whereas during the late autumn and winter macroalgae were largely absent. Previous investigations have shown that the and of benthic communities can vary on timescales of hours and days in response to the changing light conditions (Kühl et al. 2001; Glud et al. 2002). However, overall the derived P-I parameters reflect a net photosynthetic performance typical of polar micro and macroalgal species (Glud et al. 2009).

Flow velocity and O2 concentration within the BBL

The EC measurement theory invokes assumptions that the mean current velocity and O2 concentration within the BBL are at steady-state. However, transient flow velocities and O2 concentrations were evident at all four sites investigated. At site 1, rapid increases in the flow velocity magnitude of up to 15 cm s-1 over 1 or more hours coincided with a many-fold increase in the EC O2 exchange rates (Fig. 6). Transient flow velocities and O2 concentrations within the BBL can induce a temporary bias in the measured EC O2 exchange rates (Holtappels et al. 2013). To minimize inclusion of low frequency non-turbulent contributions we derived EC O2 fluxes over 10

Paper I Benthic primary production seasonality

min intervals. Although we cannot exclude that the fluxes were affected by non-turbulent contributions during certain periods, changes in the mean flow velocity and O2 concentrations typically occurred over longer timescales than this. The criteria we used to evaluate the quality of the 10 min EC O2 fluxes, in particular the rapid changes in flow direction and the ‘anisotropic ratio threshold’ method (Fig. 4) were effective at identifying periods with uncharacteristically low or high exchange rates of O2 that occur due to insufficient turbulent mixing or non-steady state conditions within the BBL. The fluxes that fell within these periods (typically amounting to <20% of each dataset) were discarded and replaced by modeled fluxes based on the P-I relation for subsequent bulk average flux estimates for NEP, R, NEM, and GPP (Table 2). Furthermore, averages derived from long term deployments such as the ones presented herein integrate much of the site-specific variability and provide trustworthy exchange rates (Holtappels et al. 2013).

Large variations between consecutive 10 min EC O2 flux estimates were common within the 18 datasets, even during periods of relatively steady flow and O2 concentrations (Figs. 4, 5). Analysis of the timeseries often indicates the presence of a few large positive and negative events within each 10 min interval in which turbulent eddies with timescales of minutes transfer O2 up and down (data not shown). The magnitude and frequency of large events within each 10 min flux interval may vary and subsequently so may the covariance estimate 〈 〉. Therefore at these timescales the EC fluxes are naturally variable. Work by McPhee (2008) has documented similar observations for under-ice heat fluxes measured by EC. The typical approach is to average the covariance estimates (in our case, 10 min) to 1 h or more to get an exchange rate that is much better representative of that taking place at the seabed surface.

Even at the hourly resolution a substantial variability in the EC fluxes was evident in our datasets

(Figs. 5-8). Detailed studies on the benthic O2 exchange rate have shown it to be highly dynamic. In

Paper I Benthic primary production seasonality

cohesive sediments or solid interfaces, the thinning of the diffusive boundary layer (DBL) during periods of enhanced flow enhances the O2 exchange (Kühl et al. 1996; Glud et al. 2007). The maximum DBL impedance of an O2 consuming sediment can be approximated by its theoretical elimination according to Boudreau and Guinasso (1982). Elimination of the 350 µm thick DBL in O2 microprofiles taken in recovered sediment cores from Nipisat Sound (site 1) during April 2012 enhanced the benthic O2 uptake in the dark by a maximum of 1.3 times, from 1.0 to 1.3 mmol O2

-2 -1 m h (data not shown). By contrast, 5 fold increases in the EC O2 fluxes with flow were common

(Fig. 6). The DBL thickness may therefore account for an appreciable but modest fraction of the variability observed within the EC O2 fluxes. For highly permeable sediments, advective pore-water flows dominate the exchange processes of O2 in the sediment surface layers and may substantially alter the benthic O2 exchange rate (Cook et al. 2007; Berg et al. 2013). Although the benthic O2 exchange rate is often seen to be highly variable in such systems, the flow-induced variations in the benthic O2 exchange rates do not necessarily scale with the actual sediment O2 turnover rates.

The variations may instead reflect the effects of an expanding oxic volume of sediment that can act as an ‘O2 reservoir’, supporting O2 consumption during periods of reduced flow, and characterized by a suppressed benthic O2 exchange rate (Cook et al 2007). Furthermore, changing pools of organic matter (e.g., photosynthesates) as well as variations in fauna activity could substantially affect O2 uptake rates during certain hours of the diel period (Fenchel and Glud 2000;

Wenzhöfer and Glud 2004). Appropriate spatial and temporal integration is therefore required to account for the site-specific natural variability, to provide robust estimates for carbon production and mineralization in coastal benthic environments (Glud 2008).

Paper I Benthic primary production seasonality

EC measurements in coastal benthic habitats

The EC method provides the opportunity to investigate complex benthic ecosystems non- invasively. The number of EC studies in coastal environments are likely to increase in the future and concurrent with the development of faster and more robust sensors (e.g., optical sensors,

Chipman et al. 2012), be expanded to resolve fluxes for a broader range of environmental parameters. The following section aims to outline some important considerations for deployment planning and interpretation of EC fluxes in coastal environments.

A main consideration in deployment planning is for the user-defined measurement height above the seabed. Generally, the position best suited for extracting EC measurements is from within the

‘constant flux’ layer within the well-mixed BBL (Brand et al. 2008), typically located 1.5-2 times above the zero plane displacement (Burba 2013). Measurements that are carried out too close to the boundary may instead characterize more localized sources due to bottom features such as stones, plants, and macrofauna, and are therefore not necessarily representative of the benthic ecosystem as a whole.

The selected measurement height has further implications for the size of the EC flux footprint (and therefore the extent to which the benthic communities are well-represented within the resolved

EC fluxes), as well as the ability to accurately correlate the EC fluxes to environmental variables

(e.g., flow velocity, PAR). Evaluating the area of the seabed that is integrated within the O2 exchange measurements is desirable to better interpret EC measurements. While this parameter is well-constrained when using benthic chambers and microprofilers, it is much less so for EC measurements. We followed the parameterization of Berg et al. (2007) to estimate the length, width, and the distance of the region that contributes most to the measured EC exchange rates, as a function of the and the measurement height above the seabed. The mean for the shallow

Paper I Benthic primary production seasonality

sites ranged from 0.001-0.017 m, resulting in footprint lengths of between 20-74 m with a width of

~1.6 m and a region of maximum flux located 0.6-2.5 m upstream from the EC instrument.

Furthermore, we followed the procedure described by Rheuban and Berg (2013) to estimate the potential errors induced due to spatial variability of the benthic communities as well as temporal changes in benthic exchange rates of O2. The uneven distribution of the flux signal within the EC measurement footprint may affect the derived EC exchange rates in heterogeneous benthic environments, such that EC exchange rates may not be representative of the total benthic community. Since the aim of our study was to obtain EC O2 exchange rates that are representative of the total benthic communities at each of our shallow sites, we evaluated whether our measurement setup was optimal for this purpose. We estimated the heterogeneity of the benthic communities from photo documentation at the four sites, and deduced that under most conditions encountered the spatial scales that were included in the EC measurements were large relative to the spatial scale of variation on the seabed. Therefore, any errors that were induced due to the spatial variability were minimal (< 5%). This analysis gives us confidence that EC O2 fluxes we present are representative of the total benthic community at each of the investigated sites.

The selected measurement height also has implications for correlating the EC O2 fluxes to environmental variables (e.g., PAR). The time required for changes in the benthic O2 exchange rate to be registered at the EC measuring height may be significantly longer than the time window over which the EC fluxes are analyzed, becoming larger as the measurement height increases and the

decreases (Rheuban and Berg 2013). The range of values (0.001-0.017m) and values

(0.004-0.023) we estimated indicate rough bed geometry typical of sands, gravels, and cobbles

(Reidenbach et al. 2010). A rough bed geometry results in vigorous turbulent mixing that rapidly

Paper I Benthic primary production seasonality

transports the signal from the seabed to the EC measurement point. As a result, the time required to transport the signal from the seabed to the EC measurement height was for most cases found to be between 10-30 min and in all cases did not exceed 50 min. This time interval is shorter than the EC O2 exchange average time windows used (i.e., 1-4 h) to derive the P-I relationships. In this respect, the P-I parameters presented in Table 2 ( , , ) therefore reflect accurately-correlated

EC O2 fluxes with PAR. Further confirmation of this was inferred by deriving P-I relations from EC fluxes averaged over time windows ranging from 10 min to several hours, which indicated that the

P-I parameters remained largely unaffected by the time-averaging window (see Fig. 4).

The high quality of data presented in this study along with the detailed inferences on the biological performance of the benthic communities that we were able to derive from these measurements gives us confidence that the EC method can provide robust benthic exchange rates in heterogeneous benthic communities under changing environmental conditions, provided that careful consideration is given to deployment planning and data analysis.

Paper I Benthic primary production seasonality

References

Barranguet C., J. Kromkamp, and J. Peene. 1998. Factors controlling primary production and photosynthetic characteristics of intertidal microphytobenthos. Mar. Ecol. Prog. Ser. 173: 117-126.

Berg, P., H. Roy, and P. L. Wiberg. 2007. Eddy correlation flux measurements: The sediment surface area that contributes to the flux. Limnol. Oceanogr. 52: 1672-1684.

Blicher, M. E., L. M. Rasmussen, M. K. Sejr, F. R. Merkel, and S. Rysgaard. 2011. Abundance and energy requirements of eiders (Somateria spp.) suggest high predation pressure on macrobenthic fauna in a key wintering habitat in SW Greenland. Polar Biology 34: 1105-1116.

Bluteau, C. E., N. L. Jones, and G. N. Ivey. 2011. Estimating turbulent kinetic energy dissipation using the inertial subrange method in environmental flows. Limnol. Oceanogr.: Methods 9: 302-321.

Boudreau, B. P., and N. L. Guinasso. 1982. The influence of a diffusive sublayer on aggregation, dissolution and diagenesis at the sea floor, p. 115-145. In K. A. Fanning and M. Manheim [eds.], The dynamic environment of the seafloor. Lexington.

Paper I Benthic primary production seasonality

Cahoon, L. B. 1999. The role of benthic microalgae in neritic ecosystems. Oceanogr. Mar. Biol. Annu. Rev. 37: 47–86.

Dalsgaard, T. 2003. Benthic primary production and nutrient cycling in sediments with benthic microalgae and transient accumulation of macroalgae. Limnol. Oceanogr. 48: 2138-2150.

Epping, E. H. G., and B. B. Jørgensen. 1996. Light-enhanced oxygen respiration in benthic phototrophic communities. Mar. Ecol. Prog. Ser. 139: 193–203.

Fenchel, T., and R. N. Glud. 2000. Benthic primary production and O2 - CO2 dynamics in a shallow-water sediment: Spatial and temporal heterogeneity. Ophelia 53: 159–171.

Paper I Benthic primary production seasonality

Glud, R. N., M. Kühl, F. Wenzhöfer, and S. Rysgaard. 2002. Benthic diatoms of a high Arctic fjord (Young Sound, NE Greenland): Importance for ecosystem primary production. Mar. Ecol. Prog. Ser. 238: 15-29.

Glud, R. N., P. Berg, H. Fossing, and B. B. Jørgensen. 2007. Effect of the diffusive boundary

layer (DBL) on the benthic mineralization and O2 distribution: A theoretical modelling exercise. Limnol. Oceanogr. 52: 547–557.

Glud, R. N. 2008. Oxygen dynamics of marine sediments. Mar. Biol. Res. 4: 243–289.

Glud, R. N., J. Woelfel, U. Karsten, M. Kuhl, and S. Rysgaard. 2009. Benthic microalgal production in the Arctic: Applied methods and status of the current database. Bot. Mar. 52: 559-678.

Jahnke, R. A., J. R. Nelson, R. L. Marinelli, and J. E. Eckman. 2000. Benthic flux of biogenic elements on the Southeastern US continental shelf: Influence of pore water advective transport and benthic microalgae. Continental Shelf Research 20: 109-127.

Jassby, A. D., and T. Platt. 1976. Mathematical formulation of relationship between photosynthesis and light for phytoplankton. Limnol. Oceanogr. 21: 540–547

Paper I Benthic primary production seasonality

Kromkamp, J., C. Barranguet, and J. Peene. 1998. Determination of microphytobenthos PSII quantum efficiency and photosynthetic activity by means of variable chlorophyll fluorescence. Mar. Ecol. Prog. Ser. 162: 45-55.

Kühl, M., R. N. Glud, J. Borum, R. Roberts, and S. Rysgaard. 2001. Photosynthetic performance of surface-associated algae below sea ice as measured with a pulse- amplitude-modulated (PAM) fluorometer and O2 microsensors. Mar. Ecol. Prog. Ser. 223: 1-14.

Long, M. H., P. Berg, D. de Beer, and J. C. Zieman. 2013. In situ coral reef oxygen metabolism: An eddy correlation study. PloS one 8: e58581, doi:10.1371/journal.pone.0058581

Lorke, A., D. F. McGinnis, and A. Maeck. 2013. Eddy-correlation measurements of benthic fluxes under complex flow conditions: Effects of coordinate transformations and averaging time scales. Limnol. Oceanogr.: Methods 11: 425-437.

Lorrai, C., D. F. McGinnis, P. Berg, A. Brand and A. Wüest. 2010. Application of oxygen eddy correlation in aquatic systems. J. Atmosph. Ocean. Tech. 27: 1533-1546.

McGee, D., R. A. Laws, and L. B. Cahoon. 2008. Live benthic diatoms from the upper continental slope: Extending the limits of marine primary production. Mar. Ecol. Prog. Ser. 356: 103-112.

McGinnis, D. F., P. Berg, A. Brand, C. Lorrai, T. J. Edmonds and A. Wüest. 2008. Measurements of eddy correlation oxygen fluxes in shallow freshwaters: Towards routine applications and analysis. Geophys. Res. Lett. 35: L04403, doi:10.1029/2007GL032747

Paper I Benthic primary production seasonality

McPhee, M. 2008. Air-ice-ocean interaction: turbulent ocean boundary layer exchange processes. Springer.

Mori, N., T. Suzuki and S. Kakuno. 2007. Noise of acoustic Doppler velocimeter data in bubbly flow. Journal of Engineering Mechanics 133: 122-125.

Reidenbach, M. A., H. Limm, M. Hondzo, and M. T. Stacey. 2010. Effects of bed roughness on boundary layer mixing and mass flux across the sediment-water interface. Water Resources Research 46: W07530, doi:10.1029/2009WR008248

Revsbech N. P. 1989. An oxygen microelectrode with a guard cathode. Limnol. Oceanogr. 34: 474-478.

Rheuban J. E., and P. Berg. 2013. The effects of spatial and temporal variability at the sediment surface on aquatic eddy correlation flux measurements. Limnol. Oceanogr.: Methods 11: 351-359.

Rysgaard, S., J. Mortensen, T. Juul-Pedersen, L. L. Sørensen, K. Lennert, D. H. Søgaard, K. E.

Arendt, M. E. Blicher, M. K. Sejr, and J. Bendtsen. 2012. High air-sea CO2 uptake rates in nearshore and shelf areas of Southern Greenland: Temporal and spatial variability. Marine Chemistry 128-129: 26-33.

Paper I Benthic primary production seasonality

Wassmann, P., C. M. Duarte, S. Agusti, and M. K. Sejr. 2011. Footprints of climate change in the Arctic marine ecosystems. Global Change Biology 17: 1235-1249.

Wenzhöfer, F., and R. N. Glud. 2004. Small-scale spatial and temporal variability in coastal

benthic O2 dynamics: Effects of fauna activity. Limnol. Oceanogr. 49: 1471-1481.

Paper I Benthic primary production seasonality

Table 1: Deployment characteristics for the 18 deployments. Values for depth, temperature, salinity, and oxygen are derived from the

CTD data and represent the mean ± standard deviation (SD) over the entire deployment.

Deployment Site Deployment Depth Temperature O O PAR Season Deployed on Salinity 2 2 No. reference length (h) (m) (°C) (μmol L-1) (% saturation) (mmol m-2 d-1) 1 1 09 May 2011 27 4.1 ± 0.8 2.5 ± 0.2 29.7 ± 0.1 379.5 ± 15.3 109.1 ± 4.8 8013.7

Spring 2 2 18 May 2011 30 5.3 ± 0.6 2.3 ± 0.3 33.1 ± 0.1 422.7 ± 10.6 123.4 ± 4.1 10508.9

3 3 25 May 2011 25 5.2 ± 0.6 2.2 ± 0.4 33.2 ± 0.1 414.8 ± 15.4 120.8 ± 5.5 5740.5

4 1 17 Jun 2011 119 4.2 ± 1.2 3.6 ± 0.5 33.3 ± 0.3 390.0 ± 27.4 117.6 ± 9.1 7059.1

5 2 30 Jun 2011 30 5.5 ± 1.2 5.3 ± 0.8 32.5 ± 0.3 405.4 ± 29.0 127.4 ± 8.7 7813.6 Summer 6 3 14 Jul 2011 35 5.3 ± 1.2 4.7 ± 0.6 32.6 ± 0.3 386.8 ± 14.7 119.8 ± 5.7 8730.8

7 1 21 Aug 2011 25 5.2 ± 1.2 5.4 ± 0.5 28.3 ± 0.6 389.9 ± 10.0 118.8 ± 3.2 8935.8

8 3 22 Sep 2011 35 5.9 ± 0.9 4.3 ± 0.2 30.1 ± 0.3 332.9 ± 3.7 100.1 ± 1.4 3672.8

9 4 04 Oct 2011 53 22.5 ± 0.7 4.4 ± 0.1 31.5 ± 0.1 330.5 ± 4.4 99.5 ± 1.5 391.9

Autumn 10 1 17 Oct 2011 30 3.1 ± 1.0 -0.1 ± 0.6 29.1 ± 0.0 380.8 ± 5.9 101.7 ± 1.1 1265.9

11 2 26 Oct 2011 50 6.8 ± 1.6 2.4 ± 0.1 31.9 ± 0.1 327.5 ± 2.5 95.2 ± 0.7 1270.9

12 3 09 Dec 2011 63 8.1 ± 1.1 0.9 ± 0.2 32.8 ± 0.1 326.8 ± 2.4 92.1 ± 0.5 102.6

13 1 20 Jan 2012 42 4.8 ± 1.1 -1.4 ± 0.4 33.1 ± 0.1 347.4 ± 2.9 92.2 ± 1.4 540.7 Winter 14 2 08 Feb 2012 50 7.0 ± 1.5 -1.3 ± 0.1 32.3 ± 0.2 349.6 ± 1.7 92.4 ± 0.3 935.5

Spring 15 1 11 Apr 2012 60 5.0 ± 1.0 0.8 ± 0.1 33.2 ± 0.1 344.0 ± 7.4 96.7 ± 2.1 8469.9

16 1 22 Jun 2012 30 4.4 ± 1.1 5.5 ± 0.9 31.7 ± 0.2 385.5 ± 7.0 120.9 ± 3.5 7433.7

Summer 17 3 27 Jun 2012 24 5.5 ± 0.8 5.0 ± 0.6 32.3 ± 0.2 368.8 ± 4.7 114.3 ± 2.3 10871.1

18 2 28 Jun 2012 24 7.1 ± 0.8 4.3 ± 0.3 32.6 ± 0.1 398.7 ± 11.6 122.3 ± 4.2 6693.0

Paper I Benthic primary production seasonality

Table 2: Eddy-correlation O2 exchange rates (mean ± standard error (SE)) for the 18 deployments.

The rates for daytime net ecosystem production (NEP), nighttime community respiration (R), and

-2 -1 benthic gross primary production (GPP) are presented in mmol O2 m h . The duration of the day and night periods are also presented. The net ecosystem metabolism (NEM) is the O2 exchange, in

-2 -1 mmol O2 m d , as weighted to a 24 h day-night period. Positive fluxes indicate a net release of O2 and negative values indicate a net uptake of O2 by the benthic ecosystems. The last four columns are the fitting parameters for the photosynthesis vs. irradiance (P-I) relationships. Values for the

-2 -1 maximum rate of gross primary production (Pm, in mmol O2 m h ), the light saturation parameter

-2 -1 -2 -1 (Ik, in μmol m s ), and the respiration component (R, in mmol O2 m h ) were determined by least-squares fitting of the P-I model to the measured EC fluxes that were averaged down from the

-2 -1 original 10 min intervals to 1 or more hours. The compensation irradiance (Ic, in μmol m s ) was subsequently inferred as the x-intercept of the P-I curve.

P-I fitting parameters Deployment Site Hours of Season Deployed on P I R I No. reference NEP R day-night NEM GPP m k c

1 1 09 May 2011 3.57 ± 0.38 -2.11 ± 0.25 16.6-7.4 43.6 5.68 6.8 116 2.3 40 Spring 2 2 18 May 2011 1.22 ± 0.12 -0.46 ± 0.06 17-7 17.5 1.68 1.9 126 0.5 32 3 3 25 May 2011 0.10 ± 0.14 -1.43 ± 0.25 18-6 -6.8 1.53 1.8 74 1.4 84 4 1 17 Jun 2011 2.02 ± 0.11 -1.73 ± 0.10 17.6-6.4 24.5 3.75 5.0 130 1.3 33

5 2 30 Jun 2011 1.05 ± 0.19 -0.81 ± 0.30 19.7-4.3 17.2 1.86 1.8 139 0.4 33 Summer 6 3 14 Jul 2011 -0.01 ± 0.08 -0.71 ± 0.16 17.5-6.5 -4.8 0.70 1.0 63 0.7 62 7 1 21 Aug 2011 1.27 ± 0.14 -2.02 ± 0.24 15.4-8.6 2.2 3.29 3.9 95 1.9 53 8 3 22 Sep 2011 0.35 ± 0.07 -0.44 ± 0.05 11-13 -1.9 0.79 1.0 63 0.4 28 9 4 04 Oct 2011 0.40 ± 0.06 -0.34 ± 0.05 8.6-15.4 -1.8 0.74 1.6 18 0.3 4 Autumn 10 1 17 Oct 2011 0.26 ± 0.05 -0.73 ± 0.10 9.4-14.6 -8.2 0.99 1.2 59 0.7 41 11 2 26 Oct 2011 0.89 ± 0.13 -0.39 ± 0.03 8-16 0.9 1.28 1.7 31 0.7 14 12 3 09 Dec 2011 -0.02 ± 0.01 -0.08 ± 0.01 6-18 -1.6 0.06 0.2 11 0.1 6 13 1 20 Jan 2012 0.15 ± 0.03 -0.19 ± 0.02 5.6-18.4 -2.7 0.34 0.5 17 0.2 7 Winter 14 2 08 Feb 2012 0.65 ± 0.07 -0.28 ± 0.03 6.8-17.2 -0.4 0.93 1.6 42 0.3 7 Spring 15 1 11 Apr 2012 0.50 ± 0.02 -0.13 ± 0.01 14.4-9.6 6.0 0.63 0.9 181 0.1 26 16 1 22 Jun 2012 1.07 ± 0.13 -0.55 ± 0.13 17.7-6.3 15.5 1.62 3.2 190 0.5 28 Summer 17 3 27 Jun 2012 -0.15 ± 0.30 -1.72 ± 0.39 18.5-5.5 -12.2 1.57 4.0 156 2.4 109 18 2 28 Jun 2012 0.30 ± 0.09 -0.60 ± 0.10 17.5-6.5 1.4 0.90 1.0 71 0.6 42

Paper I Benthic primary production seasonality

Figure legends

Fig. 1: Map indicating the location of our four study sites in Godthåbsfjord, south west Greenland.

Modified from Glud et al. (2010).

Fig. 2: The shallow sites investigated: 1) a muddy-sand embayment, 2) a sand-gravel embayment,

3) a rock-gravel embayment, and 4) a deeper muddy site. The images for sites 1-3 were obtained using a submersible camera with external strobes, while those for site 4 was taken with a small frame-mounted camera under natural light (hence the poorer quality). The image for site 1 is modified from Glud et al. (2010).

Fig. 3: The analysis carried out to investigate the effects of averaging time on the mean covariance and subsequent flux estimates for the friction velocity ( ) and the O2 fluxes. The selected time interval of 10 min is a tradeoff between including as many of the flux-contributing eddies as possible while excluding low-frequency non-turbulent contributions.

Fig. 4: The derived 10 min EC O2 exchange rates were evaluated for quality based on three criteria, namely 1) collisions of particles or debris with the O2 microsensor, 2) rapid changes in flow direction, and 3) the level of anisotropic turbulence, computed as the ratio between the horizontal and vertical dissipation rate of turbulent kinetic energy (ε). (F) Periods with low flow were typically characterized by (E) high anisotropic ratios and (C) suppressed exchange rates of O2, indicating that eddy theory may not hold during these periods. (E) The broken line (y = 5) indicates the threshold ratio used to exclude data. (A) The grey box indicates the section of O2 data that was compromised possibly due to debris hitting and getting attached to the sensor. (C) The black bars indicate the 10 min exchange rates that were excluded due to 1 or more violations in the three

Paper I Benthic primary production seasonality

screening criteria. Altogether the screening process typically resulted in <20% of the data being excluded.

Fig. 5: As a typical data example, a 5 day long eddy-correlation dataset from site 1 in June 2011

(deployment 4). (D) The eddy-correlation O2 exchange rates are presented in 10 min (grey points) and 2 h averages (bars). The 2 h averages are presented ± standard error (SE); n = 12). Gaps in the

10 min data are periods that have been excluded due to the screening process (see Methods).

Gaps that altogether constituted <20% of the data were filled with modeled fluxes based on the P-

I relationship. (A) A distinct diel cycle in the O2 exchange rates is observed in response to the available PAR. Positive O2 fluxes indicate a net O2 production (autotrophic) and negative O2 fluxes indicate a net O2 consumption (heterotrophic) benthic ecosystem. (C) The flow velocity affected the EC exchange rates during both day and night.

Fig. 6: The two main EC O2 exchange drivers were (A) the light availability and (B) the flow velocity magnitude. The P-I relationship is from a 120 h deployment from site 1 in June 2011 (deployment

4; Fig. 5). The P-I model (broken line) is fit to the 3 h flux averages using a least-squares fit.

2 Averaging of the EC O2 exchange rates markedly increased the coefficient of determination (R =

0.36 and 0.73 for the 10 min and 3 h flux averages, respectively) but did not result in significant changes in the estimated maximum photosynthetic rate ( ), the light saturation parameter ( ), or the respiration parameter ( ) that are derived from the fit. Panel B shows a typical relationship between the flow velocity and the EC O2 exchange rates during night and day at site 1

(deployment 4). A linear regression is fitted to the 10 min exchange rates to illustrate the effect of

-1 a flow increase or decrease on the EC O2 exchange rate. An increase in the flow from 2 to 12 cm s

-2 -1 2 during the day enhances the EC O2 exchange by 3.5 times, from 2.1 to 7.4 mmol O2 m h (R =

100

Paper I Benthic primary production seasonality

-1 0.43). Similarly an increase in flow from 2 to 8 cm s during the night enhances the rate of O2

-2 -1 2 exchange from -0.1 to -2.7 mmol O2 m h (R = 0.58). The closed symbols are 1 h averages ± SE (n

= 6).

Fig. 7: Diel time series of EC O2 exchange rates (presented in 1 h averages ± SE; n = 6) from each season at site 1 (spring = deployment 1, summer = deployment 4, autumn = deployment 10, winter = deployment 13). Note that the scales on the y-axes are different for the different plots to better illustrate the detail. Despite strong seasonal variances of in situ irradiance the benthic communities were autotrophic year-round during the daytime.

Fig. 8: EC O2 exchange time series under low light conditions at (A) 22 m depth during October

2011 (site 4, deployment 9) and (B) at 7 m depth during February 2012 (site 2, deployment 14).

Despite low light, the benthic communities showed high photosynthetic efficiency. Rates of

-2 -1 benthic gross primary production of up to 1.7 mmol O2 m h were measured during peak irradiance. (C) is a 6 h section from (B) deployment 14 that illustrates that little light was required to drive an autotrophic response. In this case the estimated compensation irradiance (Ic) was 2.3

μmol m-2 s-1.

Fig. 9: A time series of the photosynthetic parameters derived from the P-I relationships for the four sites. A seasonal trend in the P-I parameters was observed in response to the availability of

PAR. The compensation irradiance ( ) and light saturation ( ) values were low, indicating a benthic community able to photosynthesize efficiently under low light conditions, typical of high- latitude ecosystems.

101

Paper I Benthic primary production seasonality

Greenland

64º45’N

1 64º15’N Nipisat Sound 2 3 4 Kobbefjord

Davis Strait 10 km

51º00’W

Fig. 1

102

Paper I Benthic primary production seasonality

Site 1 Site 2

10 cm 10 cm

Site 3 Site 4

10 cm 10 cm

Fig. 2

103

Paper I Benthic primary production seasonality

Fig. 3

104

Paper I Benthic primary production seasonality

Fig. 4

105

Paper I Benthic primary production seasonality

Fig. 5

106

Paper I Benthic primary production seasonality

Fig. 6

107

Paper I Benthic primary production seasonality

Fig. 7

108

Paper I Benthic primary production seasonality

Fig. 8

109

Paper I Benthic primary production seasonality

Fig. 9

110

111

Paper II Oxygen exchange in cold-water coral communities

112

Paper II Oxygen exchange in cold-water coral communities

4.2. Paper II

Benthic O2 uptake of two cold-water coral communities estimated with the

non-invasive eddy correlation technique

L. Rovelli, K. M. Attard, L. D. Bryant, S. Flögel, H. J. Stahl, J. M. Roberts,

P. Linke, and R. N. Glud

Submitted (08/2014) to Marine Ecology Progress Series

113

Paper II Oxygen exchange in cold-water coral communities

114

Paper II Oxygen exchange in cold-water coral communities

Benthic O2 uptake of two cold-water coral communities estimated with the non-invasive eddy- correlation technique.

Lorenzo Rovelli, 1,2* Karl M. Attard, 2,3 Lee D. Bryant, 5,6 Sascha Flögel, 5 Henrik J. Stahl, 2 J. Murray Roberts, 7,1,8 Peter Linke, 5 and Ronnie N. Glud, 1,2,3,4

1 Scottish Association for Marine Sciences, Scottish Marine Institute, PA37 1QA Oban, United Kingdom

2 University of Southern Denmark, Nordic Centre for Earth Evolution (NordCEE), 5230 Odense M, Denmark

3 Greenland Institute of Natural Resources, Greenland Climate Research Centre, 3900 Nuuk, Greenland

4 University of Århus, Arctic Research Centre, 8000 Århus C, Denmark

5 Helmholtz Centre for Ocean Research Kiel (GEOMAR), 24148 Kiel, Germany

6 now at University of Bath, Department of Architecture and Civil Engineering, BA2 7AY Bath, United Kingdom

7 Heriot-Watt University, Centre for Marine Biodiversity and Biotechnology, School of Life Sciences, EH14 4AS Edinburgh, United Kingdom

8 University of North Carolina Wilmington, Center for Marine Science, 28409 Wilmington, United States of America

* Corresponding author: [email protected]

Keywords: eddy correlation, cold-water coral, community oxygen exchange, Mingulay Reef Complex, Stjernsund

Running title: Oxygen exchange in cold-water coral communities

115

Paper II Oxygen exchange in cold-water coral communities

Acknowledgements

We are grateful to the crew of the RRS James Cook and R/V Poseidon for their assistance throughout the respective research cruises. We thank A. Glud for proving in the O2 electrodes used in this study, A. Lorke for the Matlab ID script, as well as R. Schwarz and S. Cherednichenko of

GEOMAR’s Technology and Logistic center for their support during designing and building of the

ECM frame, amplifiers and timer module as well as during the preparation and programming at sea. Furthermore, we thank the Holland-1 and Phoca ROV teams for safe deployment and recovery at the Mingulay and Stjernsund sites respectively. Funding for the Mingulay deployment was provided through the UK Ocean Acidification programme (NERC grant NE/H017305/1 and added-value awards to JMR). Funding for Molab was provided by the Federal Ministry of

Education and Research (BMBF) under grant 03F06241; the Poseidon cruises POS434 and POS438 were supported by GEOMAR and industry funding (grant A2300414 to PL). LR, KMA & RNG, received financial support from National Environmental Research Council (NERC) – NE/F018614/1;

NE/F0122991/1: The Commission for Scientific Research in Greenland (KVUG) – GCRC6507; The

Danish Council for Independent Research (FNU-12-125843); ERC Advanced Grant, ERC-2010-

AdG_20100224 and The Danish National Research Foundation (DRNF53)

116

Paper II Oxygen exchange in cold-water coral communities

Abstract

To assess their community respiration, benthic O2 uptake at two contrasting tidally-dominated cold-water coral (CWC) sites was estimated using the non-invasive eddy correlation (EC) technique. The first site, Mingulay Reef Complex, was a rock ridge located in the Sea of Hebrides off Scotland at a depth of 138 m and the second site, Stjernsund, was a channel-like sound in

Northern Norway at a depth of 220 m. Both sites were characterized by the presence of live mounds of the reef framework-forming scleractinian Lophelia pertusa and reef-associated fauna such as sponges, crustaceans and other corals. The measured O2 uptake at the two sites varied between -5 and -46 mmol m-2 d-1, mainly depending on the ambient flow characteristics. The average value estimated from the ~24 h long deployments amounted to -27.6 ± 2.0 mmol m-2 d-1 at Mingulay and -24.5 ± 1.2 mmol m-2 d-1 at Stjernsund (mean ± standard error), respectively.

These rates are four to five times higher than the global mean for soft sediment communities at comparable depths. The measurements document the importance of CWC communities for local and regional carbon cycling and demonstrate that the EC technique is a valuable tool for assessing rates of benthic O2 uptake in such complex and dynamic settings.

117

Paper II Oxygen exchange in cold-water coral communities

118

Paper II Oxygen exchange in cold-water coral communities

Introduction

Cold-water corals (CWC) are azooxanthellate cnidarians that occur worldwide on continental shelves, slopes, seamounts, and ridge systems. CWC may occur as individual polyps (such as the cup corals), as discrete colonies, such as the black corals (Antipatharia), or as reef framework- forming colonies such as the scleractinian Lophelia pertusa. The distribution may extend anywhere from small patches on the seabed to giant coral carbonate mounds of several kilometers in diameter (Roberts et al. 2006). The complex structural framework provides niches for a great biomass and diversity of organisms such as, echinoderms, crustaceans, and sponges and fish

(Roberts & Cairns 2014). The communities trap considerable amounts of organic material and is expected to turnover considerable amounts of organic material (van Oevelen et al. 2009, White et al. 2012). This has emphasized the need for more studies to further investigate CWC metabolism and to constrain CWC community contributions to local and regional carbon budgets (Roberts et al. 2009).

However, the assessment of CWC reef community respiration rates poses major methodological challenges due to the structural complexity and spatial heterogeneity of the reefs. Traditional methods are invasive and rely on ex-situ incubations of reef coral fragments (Dodds et al. 2007), in-situ incubations of reef community sub-samples combined with complex food-web modeling

(van Oevelen et al. 2009), and on in-situ benthic chamber deployments (Khripounoff et al. 2014) which exclude the natural hydrodynamics, and are difficult to apply to many reef areas. Currently, non-invasive integrated estimates of CWC reef community respiration rates are limited to open water approaches that rely on benthic boundary layer (BBL) O2 budget estimates from two or more coincident fixed-point O2 measurements over the selected CWC reef area (White et al.

119

Paper II Oxygen exchange in cold-water coral communities

2012). The robustness of the O2 budget strongly depends on how well the water mass residence time within the target area is constrained spatially and temporally and therefore appears most suitable for channel-like sites.

The O2 eddy-correlation (EC) technique is a non-invasive in-situ technique that is used to infer the benthic O2 exchange rate from direct Eulerian measurements of the turbulent transport of O2 within the BBL (Berg et al. 2003). The technique has been successfully applied to structurally complex benthic surfaces such as tropical coral reefs (Long et al. 2013), high-latitude rocky embayments (Glud et al. 2010) and oyster beds (Reidenbach et al. 2013). Since EC O2 exchange measurements are extracted under the natural hydrodynamics and from large areas of the seabed, the measurements may integrate much of the site-specific spatial and temporal variability to provide robust measures of benthic O2 exchange of heterogeneous benthic communities

(Rheuban & Berg 2013).

Here, we apply the EC technique to assess the integrated community respiration of two contrasting settings: 1) the Mingulay Reef Complex located in the Sea of Hebrides off Scotland, and 2) the Stjernsund located in northern Norway.

Methods

Study sites

Mingulay Reef Complex

The Mingulay Reef Complex (MRC, 56º 47’ N – 7º 25 E; Fig.1a) is a 20-km long, 10-km wide, 70 –

250 m deep area located in the Sea of the Hebrides off Scotland characterized by the localized occurrence of CWC structures (Eden et al. 1971, Griffiths 2002, Roberts et al. 2005). MRC seabed

120

Paper II Oxygen exchange in cold-water coral communities

topography comprised of hollows and rock ridges and is the result of a North–South graben structure of glacial origin and successive sedimentation (Roberts et al. 2009). This study was performed at Mingulay Area 1 (see Roberts et al. 2009 and references therein) during the

“Changing Oceans” cruise, RRS James Cook cruise 073 (18 May – 15 June 2012). The EC instrument was placed on the seabed by remotely operated vehicle (ROV) Holland-1 (see: https://www.marine.ie/home/services/researchvessels/ROV.htm) at a depth of 138 m. Small coral mounds of live L. pertusa as well as crinoids (Leptometra sp.) and sponges were present nearby the instrument (Fig. 1). The flow velocity at the seabed was tidally driven and ranged from 5 – 30 cm s-1.

Stjernsund reef

Stjernsund (70.5 ºN – 22.5 ºE; Fig.1a) is a 30-km long, 3.5-km wide, >400-m deep glacial sound in northern Norway that connects the Altafjord to North-Atlantic waters (Rüggeberg et al. 2011). A submerged SW-NE morainic sill hosts one of Europe’s northernmost CWC reefs (Dons 1932,

Freiwald et al. 1997). Targeted manned submersible inspections of the sill have identified zones at

235 – 305 m and 245 – 280 m for the NE and SW respectively, with patches of L. pertusa

(Rüggeberg et al. 2011). The crest of the sill was dominated by a sponge-hydroid community, mainly Mycale sp. and Tubularia sp., and smaller patches of living L. pertusa (Fig. 1b). The sill crest, at a depth of 220 m, was investigated in this EC study as part of a large Modular multidisciplinary seafloor observatory (Molab) deployment in Stjernsund. The Molab EC modules (ECMs) were deployed by ROV Phoca (see: http://www.geomar.de/en/centre/central- facilities/tlz/rovphoca/overview/) during the R/V Poseidon POS434 cruise (26 May – 15 June 2012) and retrieved four months later during the R/V Poseidon POS438 cruise (10 – 27 September 2012).

121

Paper II Oxygen exchange in cold-water coral communities

The flow velocity at the seabed was tidally driven and, during the observational period, ranged from 1 – 15 cm s-1.

Eddy correlation measurements

Site selection

Before the EC deployment, ROV surveys were performed in the target areas to identify suitable spots for EC measurements. The main criteria for site selection were: 1) the presence of a flat area to safely deploy the frame, 2) a sufficient distance from large meter-sized structures and features

(i.e., reef mounds and dense CWC patches), and 3) the occurrence of an established benthic community. Further emphasis was given to the EC frame orientation in respect to the local flow hydrodynamics, aiming at both deploying the ECM in line with the main flow direction and avoiding flow disturbances due to large structures in the vicinity of the ECMs. The ROV video footage was also used to describe the benthic community, i.e. the main taxa occurrence and their relative spatial coverage, within the estimated EC footprint.

Instrumental setup

The O2 flux estimations were performed with two eddy correlation modules (ECM), one from SDU

(University of Southern Denmark; Mingulay site) and one from GEOMAR (Stjernsund site). The setup of the applied ECMs was similar to the original design by Berg et al. (2003). The main components of the ECMs consisted of an acoustic Doppler velocimeter (ADV; Vector, Nortek A/S) and Clark-type O2 electrodes (Revsbech 1989) that relayed the signal to the ADV via submersible amplifiers (McGinnis et al. 2011). The O2 electrodes 90% response time was 0.5 s while the stirring sensitivity was below 0.5% (see Gundersen et al. 1998). The ADV recorded the velocity

122

Paper II Oxygen exchange in cold-water coral communities

components as well as the O2 microsensor data at a frequency of 64 Hz and in addition collected ancillary information such as the sampling distance from the seabed, flow direction, and signal strength. The EC system was mounted onto a small stainless steel tripod frame that was designed specifically for deployment by ROV (McGinnis et al. 2011). A small conductivity-temperature-depth

(CTD) logger equipped with an O2 Aanderaa optode was mounted onto each ECM to collect background environmental information and for in-situ calibration of the Clark-type O2 electrodes.

The ADV was mounted downward-facing perpendicular to the seabed surface and the ADV sampling volume was located 25 cm and 15 cm above the reefbed for the Mingulay Reef and

Stjernsund Reef deployments, respectively. For the four-month deployment at Stjernsund Reef, a timer was added to the ADV to control the instrument on – off times and thus increase battery duration in order to maintain recording over the whole deployment time. ECM timer was programmed to collect a 25 h EC measurement dataset every week. However, due to issues with the internal ADV logger, only one dataset of 22.5 h was recorded during the four months measurements period.

Data processing

The ADV 64-Hz datasets were averaged down to 8 Hz for further processing. Quality controls included flagging ADV velocity data with beam correlations <50% and signal-to-noise ratios (SNR)

<10, as well as subsequent despiking of the O2 and velocity time series (Matlab Despiking Toolbox;

Goring & Nikora 2002, Mori et al. 2007). A planar-fit coordinate rotation was performed on the 8-

Hz velocity data to obtain a vertical velocity component normal to the local streamline (Lorke et al.

2013).

123

Paper II Oxygen exchange in cold-water coral communities

The time-averaged turbulent O2 flux ( ) was estimated from vertical velocity fluctuations (

̅̅̅̅̅̅ and O2 concentration fluctuations ( ) as (Berg et al. 2003). The fluctuations were obtained via linear detrending based on Taylor decomposition as ̅ and ̅ with w and C being the measured vertical velocities and O2 concentrations, and ̅ and ̅ the time averaged values. Data averaging, time-shifting and O2 flux estimates were performed using the

Fortran program suite Sulfide-Oxygen-Heat Flux Eddy Analysis (SOHFEA) version 2.0 (available from www.dfmcginnis.com/SOHFEA; McGinnis et al. 2011). The window size (time interval) for the estimation of the turbulent fluctuations was inferred from the bulk averages over incrementally increased window sizes for O2 fluxes (McGinnis et al. 2008, Attard et al. 2014) and shear velocity

(u*; McPhee 2008). For both sites a windows size of 3 min was found to be an optimal trade-off between including the major turbulent flux contributions while minimizing the inclusion of non- turbulent processes; the resulting O2 fluxes were subsequently averaged to 1-h intervals. The representative averages for each window size were computed from Reynolds stress as

√ ̅̅ ̅ ̅̅̅ with representing longitudinal flow fluctuations (Inoue et al. 2011, Reidenbach et al. 2006). The sediment surface roughness parameter (z0) was estimated assuming logarithmic

law-of-the-wall scaling as , where is the measuring height above the

benthic surface, ҡ is the von Karman constant (0.41), and is the flow velocity magnitude (Wüest

& Lorke 2003). The mean z0 was subsequently used to estimate the characteristics of the EC flux footprint from empirical relations that evaluate the downstream transport and dispersion of a dissolved conservative tracer at the seabed (see Berg et al. 2007).

Quality refinement of O2 fluxes encompassed: 1) removal of spikes due to sensor collisions with particles and debris 2) flagging of possible directional flow dependencies e.g. during abrupt flow

124

Paper II Oxygen exchange in cold-water coral communities

direction changes, and 3) anisotropy level. The latter is based on the ratio between the average horizontal (εy) and vertical (εz) turbulent kinetic energy dissipation rates that were estimated from the ADV velocity time series using the inertial dissipation method (ID; Bluteau et al. 2011, Inoue et al. 2011). A ratio of 1 implies full isotropy, i.e. well-developed turbulence in all directions, with increasing anisotropy the larger the ratio, i.e., the stronger the directional component in the turbulence. The threshold anisotropy used to constrain EC-favorable, near-isotropic conditions is dependent on the local hydrodynamics and topography; ratios of 12 and 9 were found to be most suitable for the settings at Mingulay and Stjernsund, respectively.

Results

At MRC the EC deployment lasted 25 h. During the observational period the average temperature was 9.30 ± 0.01 ºC (mean ± standard deviation, SD) while the O2 concentration also showed minimal variation ranging from 245 to 246 µmol L-1 (88% saturation; Table 1). In agreement with field surveys (Davies et al. 2009) and subsequent modelling (Navas et al. 2014) a semi-diurnal trend was clearly observed in the ADV hydrostatic pressure dataset (Fig 2a). The trend was less apparent in the flow measurements where the observed flow variability at the hourly scale suggested more

-2 -1 complex hydrodynamics (Fig. 2c). Hourly O2 flux averages ranged from -14 to -46 mmol m d with an overall deployment average of -27.6 ± 2.0 mmol m-2 d-1 (mean ± standard error, SE; Fig. 2e).

At the Stjernsund reef, the ECM collected 22.5 h of consecutive data. During the measurements, O2 concentration ranged from 251 to 261 µmol L-1, with near-constant temperature conditions (6.10 ±

0.02; Table 1). Semi-diurnal tidal signatures were observed in the hydrostatic pressure and flow regime (Fig. 2f, h), as well as in O2 concentration (Fig. 2g), indicating a well-established tidal front moving along the Stjernsund channel. The most consistent O2 fluxes were obtained during periods

125

Paper II Oxygen exchange in cold-water coral communities

of incoming tide (Fig 2f, j; red and black boxes). In those periods the O2 concentrations remained

-1 fairly constant around 258 – 259 µmol L (86.5% saturation), and the hourly O2 fluxes ranged from

-4.8 to -40.8 mmol m-2 d-1, with an overall average -24.5 ± 1.2 mmol m-2 d-1 (mean ± SE).

At Mingulay, the EC O2 exchange rates integrated a footprint length of 11.8 m and width of 1.6 m with major flux contributions about 0.1 m from the ADV. At the Stjernsund, the footprint had a length of 3.5 m and a width of 0.9 m with the region of maximum flux located at the position of the sampling volume (Berg et al. 2007; Table 1).

At Mingulay, ROV footage showed that the deployment area was characterized by abundant fragments of dead coral rubble and small fragments of dead coral framework. The latter were frequently colonized by the zoanthid Parazoanthus anguicomus, the sponge Spongosorites coralliophaga with a prominent coverage of hydroids and the suspension feeding sabellid polychaetes. Occasional mobile invertebrates, notably the edible crab Cancer pagurus, were seen in the area. Although nearby, live L. pertusa colonies were outside the EC footprint (Fig 1c). ROV video footage analysis of the Stjernsund EC footprint revealed the occurrence of live orange L. pertusa covering the left side of the footprint area, with the intercalated presence of Mycale lingua sponges. Sparse red and white gorgonians (Paragorgia arborea) were also present in the footprint, representing, together with L. pertusa mounds, the tallest features observed with a height of up to 50 cm. Sea anemones such as Protanthea simplex and hydroids such as Tubularia indivisa covered the coral rubble that made up a large majority of the footprint (Fig 1d).

Discussion

We were able to derive robust in-situ estimates of the benthic O2 exchange rate at two contrasting

CWC reef sites using the EC method. Despite the geographic separation and the differences in

126

Paper II Oxygen exchange in cold-water coral communities

benthic communities (e.g. lack of gorgonian corals at Mingulay compared to Stjernsund) at the two measurement sites, the average O2 flux estimates were similar. The fluxes were within the wide range reported in previous CWC reef studies using other less well-constrained approaches.

-2 -1 White et al. (2012) reported benthic O2 exchange rates ranging from 10.3 to 88 mmol m d based

-2 -1 on an open-water approach. Incubation-based O2 exchange rates ranged from 7.7 mmol m d for

L. pertusa alone (Khripounoff et al. 2014) to 74.5 mmol m-2 d-1 for a whole CWC community (van

Oevelen et al. 2009; using a 138:106 O2 to C Redfield ratio). When compared to other marine systems, the reported O2 fluxes are four to five times higher than the global average total benthic

O2 uptake rate (TOU) for soft sediments derived from Glud (2008) for comparable depths (7.4 and

-2 -1 5.2 mmol O2 m d for Mingulay and Stjernsund, respectively). Similarly, those values are ~4 times higher than in-situ microprofile measurements conducted in soft muddy sediments at 329 m depth very close to Stjernsund (Van Mijen 69º 29.4’ N, 18º 07.5’ W; Glud et al. 1998).

Eddy correlation O2 fluxes are valuable for CWC studies as they are extracted under the naturally varying in-situ conditions and integrate a large footprint area. The robustness of the O2 fluxes derived depends not only upon the dataset quality, but also on a careful characterization of the footprint area with regard to size, patchiness and benthic community structure. The estimated footprint areas in this study were relatively small: within the 10 m2 range at the Mingulay site and limited to a few m2 at the Stjernsund site (Table 1). This was attributed to the natural roughness of

CWC reefs as elevated z0 results in increased turbulent mixing and thus a smaller footprint than in systems with comparable flow velocity but less rough topography. In this study z0 ranged from 2.7

– 3.6 cm (Table 1) as compared for example with rough gravel or cobble beds where z0 is typically

0.1 – 0.5 cm (Reidenbach et al. 2010). While the spatial (horizontal) coverage of the EC measurements in these heterogenic environments is larger than typical benthic chamber and

127

Paper II Oxygen exchange in cold-water coral communities

incubation experiments, one needs to consider how well that heterogeneity is integrated within the EC O2 fluxes. The effect of spatial heterogeneity on O2 fluxes has been investigated by

Rheuban & Berg (2013) based on model simulations of idealized patches and patch heterogeneity.

If we apply the patch size equations provided by the authors to the measuring heights and z0 of this study, we find that our measurements integrated approximated patch sizes in the 30 – 50 cm range at the Mingulay site and 10 – 20 cm at the Stjernsund site. Based on ROV video and stills footage (Fig. 1), the typical heterogeneity patch size with the footprint was inferred to be in the order of 10 – 30 cm and 20 – 30 cm for the Mingulay and Stjernsund site, respectively.

The average height of the observed benthic features close to the measuring site at Mingulay were estimated to be up to 20 cm (Fig. 1) and thus shorter or comparable with the EC measuring height.

However, at the Stjernsund some of the L. pertusa patches and isolated branches grew above the

EC measuring height and thus might have not been completely integrated in the measurements.

To date there remain few studies on the contribution of L. pertusa respiration to total CWC reef community metabolism. Coupled modelling and ex-situ incubations from 800-m deep northeast

Atlantic Rockall Bank coral carbonate mounds by van Oevelen et al. (2009) have indicated that L. pertusa only contributes 9% of the total CWC reef community O2 consumption. These results agree with the recent study of Khripounoff et al. (2014) who showed that, at a broadly comparable site, a CWC density of three live coral branches per m2 contribute to ~8 mmol m2 d-1. Therefore, although it is possible that some contributions from the L. pertusa mounds within the footprint were excluded from the derived rate estimates, it is likely that including missing portions of the mounds would not substantially alter the rates presented here.

128

Paper II Oxygen exchange in cold-water coral communities

Dynamic in-situ conditions such as changes in temperature (Dodds et al. 2007) and hydrodynamics

(Davies et al. 2009, Navas et al. 2014) may affect CWC reef metabolism and therefore should be considered in the measurements. Compared to traditional methods such as chamber incubations, the EC method holds clear advantages in this respect. We demonstrate that the EC method can be used to derive benthic O2 exchange rates from contrasting CWC sites with complex topography from large areas of the seabed under different environmental conditions. The results also suggest that future studies should be ideally performed higher above the seabed to integrate even more of the CWC reef’s inherent topographic heterogeneity.

Furthermore, the EC measurements may integrate mixed benthic communities provided that careful consideration is given to deployment planning and benthic community analyses through

ROV video footage or other means. These measurements provide valuable information on the ecological functioning of CWC ecosystems and may be used to investigate subsequent open questions such as the assessment of seasonal variability of CWC reef metabolism and the role of

CWC reefs in regional and global carbon cycling.

129

Paper II Oxygen exchange in cold-water coral communities

130

Paper II Oxygen exchange in cold-water coral communities

References

Attard KM, Glud RN, McGinnis DF, Rysgaard S (2014) Seasonal rates of benthic primary production in a Greenland fjord measured by aquatic eddy-correlation. Limnol Oceanogr. 59: 1555–1569, doi:10.4319/lo.2014.59.5.1555

Berg P, Roy H, Janssen F, Meyer V, Jorgensen BB, Huettel M, D. de Beer D (2003). Oxygen uptake by aquatic sediments measured with a novel non-invasive eddy-correlation technique. Mar Ecol Progr Ser 261:75–83, doi:10.3354/meps261075

Berg P, Roy H, Wiberg PL (2007) Eddy correlation flux measurements: The sediment surface area that contributes to the flux. Limnol Oceanogr 52:1672–1684, doi:10.4319/lo.2007.52.4.1672

Bluteau, CE, Jones NL, Ivey GN (2011) Estimating turbulent kinetic energy dissipation using the inertial subrange method in environmental flows. Limnol Oceanogr: Methods 9:302–321, doi:10.4319/Lom.2011.9.302

Davies AJ, Duineveld GCA, Lavaleye MSS, Bergman MJN, van Haren H, Roberts JM (2009) Downwelling and deep-water bottom currents as food supply mechanisms to the cold- water coral Lophelia pertusa (Scleractinia) at the Mingulay Reef complex. Limnol Oceanogr 54:620–629, doi:10.4319/Lo.2009.54.2.0620

Dodds LA, Roberts JM, Taylor AC, Marubini F (2007) Metabolic tolerance of the cold-water coral Lophelia pertusa (Scleractinia) to temperature and dissolved oxygen change. J Exp Mar Biol Ecol 349:205–214, doi:10.1016/J.Jembe.2007.05.013

Dons C (1932) Zoologiske Notiser XV. Om Nord-Norges korallsamfund. Det Kongelige Norske Videnskabers Selskab Forhandlingar 5(4):13–16.

Eden RA, Ardus DA, Binss PE, McQuillin R, Wilson JB (1971) Geological investigations with a manned submersible off the west coast of Scotland 1969–1970. Institute of Geological Sciences Report 71/16, p 49

131

Paper II Oxygen exchange in cold-water coral communities

Freiwald A, Henrich R, Pätzold J (1997) Anatomy of a deep-water coral reef mound from Stjernsund, West-Finnmark, northern Norway. Cool-Water Carbonates. In: James NP (ed), SEPM, Special Publication, 56, p 141–161

Glud RN (2008) Oxygen dynamics in marine sediments. Mar Biol Res 4:243–289, doi:10.1080/17451000801888726

Glud RN, Holby O, Hoffmann F, Canfield DE (1998) Benthic mineralization and exchange in Arctic sediments (Svalbard, Norway). Mar Ecol Progr Ser 173:237–251, doi:10.3354/Meps173237

Glud RN, Berg P, Hume A, Batty P, Blicher ME, Lennert K, Rysgaard S (2010) Benthic O2 exchange across hard-bottom substrates quantified by eddy correlation in a sub-Arctic fjord. Mar Ecol Progr Ser 417:1–12, doi:10.3354/Meps08795

Goring DG, Nikora VI (2002) Despiking Acoustic Doppler Velocimeter Data. J Hydraul Eng 128:117–126, doi:10.1061/(ASCE)0733-9429(2002)128:1(117)

Griffiths C (2002) Discovery 257. SAMS Northern Seas Programme, 21st September–9th October 2001. Clyde-Clyde, Internal Cruise Report, Scottish Association for Marine Science, Oban, p 54

Gundersen JK, Ramsing NB, N. Glud RN (1998) Predicting the signal of O2 microsensors from

physical dimensions, temperature, salinity, and O2 concentration. Limnol Oceanogr 43:1932–1937.

Inoue T, Glud RN, Stahl H, Hume A (2011) Comparison of three different methods for assessing in situ friction velocity: A case study from Loch Etive, Scotland. Limnol Oceanogr: Methods 9:275–287, doi:10.4319/lom.2011.9.275

Khripounoff A, Caprais J-C, Le Bruchec J, Rodier P, Noel P, Cathalot C (2014) Deep cold-water coral ecosystems in the Brittany submarine canyons (Northeast Atlantic): Hydrodynamics, particle supply, respiration, and carbon cycling. Linmnol Oceanogr 59:87–98, doi:10.4319/lo.2014.59.01.0087

132

Paper II Oxygen exchange in cold-water coral communities

Long MH, Berg P, de Beer D, Zieman JC (2013) In Situ Coral Reef Oxygen Metabolism: An Eddy Correlation Study. Plos One 8:e58581, doi:10.1371/journal.pone.0058581

Lorke A, McGinnis DF, and A. Maeck A (2013) Eddy-correlation measurements of benthic fluxes under complex flow conditions: Effects of coordinate transformations and averaging time scales. Limnol Oceanogr: Methods 11:425–437, doi:10.4319/Lom.2013.11.425

McPhee M (2008) Air-Ice-Ocean Interaction: turbulent ocean boundary layer exchange processes. Springer-Verlag, Berlin, p. 215

McGinnis DF, Berg P, Brand A, Lorrai C, Edmonds TJ, Wuest A 2008. Measurements of eddy correlation oxygen fluxes in shallow freshwaters: Towards routine applications and analysis. Geophys Res Lett 35:L04403, doi:10.1029/2007GL032747

McGinnis DF, Cherednichenko S, Sommer S, Berg P, Rovelli L, Schwarz R, Glud RN, Linke P (2011) Simple, robust eddy correlation amplifier for aquatic dissolved oxygen and hydrogen sulfide flux measurements. Limnol Oceanogr: Methods 9:340–347, doi:10.4319/lom.2011.9.340

Navas JM, Miller PL, Henry L-A, Hennige SJ, Roberts JM (2014) Ecohydrodynamics of Cold- Water Coral Reefs: A Case Study of the Mingulay Reef Complex (Western Scotland). PLoS ONE 9: e98218, doi:10.1371/journal.pone.0098218

Mori N, Suzuki T, Kakuno S (2007) Noise of acoustic Doppler velocimeter data in bubbly flow. J Eng Mech 133:122–125, doi:10.1061/(ASCE)0733-9399(2007)133:1(122)

Reidenbach MA, Monismith SG, Koseff JR, Yahel G, Genin A (2006) Boundary layer turbulence and flow structure over a fringing coral reef. Limnol Oceanogr 51: 1956– 1968, doi:10.4319/lo.2006.51.5.1956

Reidenbach MA, Limm M, Hondzo M, Stacey MT (2010) Effects of bed roughness on boundary layer mixing and mass flux across the sediment-water interface. Water Resour Res 46:W07530, doi:10.1029/2009wr008248.

133

Paper II Oxygen exchange in cold-water coral communities

Reidenbach MA, Berg P, Hume A, Hansen JCR, Whitman ER (2013) Hydrodynamics of intertidal oyster reefs: the influence of boundary layer flow processes on sediment and oxygen exchange. Limnol Oceanogr: Fluids and Environments 3:225–239, doi:10.1215/21573689-2395266

Rheuban J, Berg P (2013) The effect of benthic surface heterogeneity on eddy correlation flux measurements. Limnol Oceanogr: Methods 11:351–359, doi:10.4319/lom.2013.11.351

Revsbech NP (1989) An oxygen microelectrode with a guard cathode. Limnol Oceanogr 34:474–478, doi:10.4319/lo.1989.34.2.0474

Roberts JM, Brown CJ, Long D, Bates CR (2005) Acoustic mapping using a multibeam echosounder reveals cold-water coral reefs and surrounding habitats. Coral Reefs 24:654–669, 10.1007/S00338-005-0049-6

Roberts JM, Wheeler AJ, Freiwald A (2006) Reefs of the deep: the biology and geology of cold-water ecosystems. Science 312:543–547, doi:10.1126/science.1119861

Roberts JM, Cairns SD (2014) Cold-water corals in a changing ocean. Curr Opin Environ Sustainability 7:118–126, doi:10.1016/j.cosust.2014.01.004

Roberts JM, Davies AJ, Henry LA, Dodds LA, Duineveld GCA, Lavaleye MSS, Maier C, van Soest RWM, Bergman MJN, Huhnerbach V, Huvenne VAI, Sinclair DJ, Watmough T, Long D, Green SL, van Haren H (2009) Mingulay reef complex: an interdisciplinary study of cold-water coral habitat, hydrography and biodiversity. Mar Ecol Progr Ser 397:139–151, doi:10.3354/Meps08112

Rüggeberg A, Flögel S, Dullo WC, Hissmann K, Freiwald A (2011) Water mass characteristics and sill dynamics in a subpolar cold-water coral reef setting at Stjernsund, northern Norway. Mar Geol 282:5–12, doi:10.1016/J.Margeo.2010.05.009

van Oevelen D, Duineveld G, Lavaleye M (2009) The cold-water coral community as a hot spot for carbon cycling on continental margins: a food-web analysis from Rockall Bank (northeast Atlantic). Limnol Oceanogr 54:1829−1844, doi: 10.4319/lo.2009.54.6.1829

134

Paper II Oxygen exchange in cold-water coral communities

White M, Wolff GA, Lundalv T, Guihen D, Kiriakoulakis K, Lavaleye M, Duineveld G (2012) Cold-water coral ecosystem (Tisler Reef, Norwegian Shelf) may be a hotspot for carbon cycling. Mar Ecol Progr Ser 465:11–23, doi:10.3354/Meps09888

Wuest A, Lorke A (2003) Small-scale hydrodynamics in lakes. Annu Rev Fluid Mech 35:373– 412, doi:10.1146/annurev.fluid.35.101101.161220

135

Paper II Oxygen exchange in cold-water coral communities

136

Paper II Oxygen exchange in cold-water coral communities

Figure captions

Fig. 1. Eddy Correlation (EC) module (ECM) system deployments. Location of the Mingulay reef complex, MRC (a) and Stjernsund site (b). (c, d) Close-up of the CWC communities within the EC footprint area at Mingulay reef complex (c; courtesy of the ROV Holland-1 team) and Stjernsund

(d; courtesy of the ROV Phoca team).

Fig. 2. Eddy Correlation (EC) datasets from Mingulay (panels a – e) and the Stjernsund (panels f – j) cold-water coral sites. (a, f) Hydrostatic pressure. (b, g) Bulk water O2 concentration from CTD optodes. (c, h) Flow velocities along the ADV x-axis (black lines) and y-axis (gray lines). Negative values indicate movements towards the ADV. Red dotted lines indicate current velocity magnitude. (d, i) Isotropic level, defined as the ratio between the horizontal (εz) and vertical (εz) turbulence level estimated using the inertial dissipation method. (e, j) Turbulent O2 fluxes average over 1-h intervals with associated standard error. Gray dotted fluxes indicate data were flagged due to high isotropic level, unsuitable current directions and O2 electrode collisions. Note that the time, in hours, refers to the time elapsed starting from midnight of the deployment day (e.g. 12 represents noon).

137

Paper II Oxygen exchange in cold-water coral communities

Tables

Table 1. Eddy correlation module deployments summary

Mingulay Stjernsund Location 56º 49.354’ N 70º 16.256’ N 7º 23.723’ W 22º 28.321’ E Depth (m) 138 220 Deployment duration (h) 25 22.5 Temperature (mean ± SD; ºC) 9.3 ± 0.01 6.1 ± 0.02 -1 I O2 (mean ± SD; µmol L ) 245.8 ± 0.004 [245 – 255 ± 2.6 [251 – 246] 261] Distance from bottom (h; cm) 25 13.5 Flow (cm s-1) 15.9 – 17.4 3.6 – 4.8 Shear velocity (u*; cm s-1) 3.3 – 3.4 1.02 – 1.04

Roughness parameter ( ; cm) 3.2 – 3.6 2.7 – 3.3 FootprintII Length/Width/Maximum flux (m) 11.8/1.6/0.1 3.5/0.9/0 AreaIII (m2) 15.2 2.4 IV O2 Flux -27.6 ± 2.0 (18) -24.5 ± 1.2 (19) (mean ± SE (n); mmol m-2 d-1)

I: Average over the whole observational period. Values within square brackets indicate the O2 range.

II: Based on average z0.

III: For simplicity, the footprint area was approximated to be an ellipse.

IV: Averaged values based on 18 h and 19 h of usable hourly O2 fluxes for Mingulay and Stjernsund, respectively.

138

Paper II Oxygen exchange in cold-water coral communities

Figures

Fig. 1

139

Paper II Oxygen exchange in cold-water coral communities

Fig. 2

140

141

Paper III Benthic O2 exchange in maerl and sand

142

Paper III Benthic O2 exchange in maerl and sand

4.3. Paper III

Benthic O2 exchange in a pristine maerl bed and a nearby sandy habitat in a

temperate sea loch (Loch Sween, Scotland): A seasonal, in situ study

K. M. Attard, H. Ståhl, N. A. Kamenos, G. Turner, H. L. Burdett, and R. N. Glud

In preparation for Limnology and Oceanography

Photos: Rob Cook, Heriot-Watt University

143

Paper III Benthic O2 exchange in maerl and sand

144

Paper III Benthic O2 exchange in maerl and sand

Benthic O2 exchange in a pristine maerl bed and a nearby sandy habitat in a temperate sea loch

(Loch Sween, Scotland): A seasonal, in situ study

Karl M. Attard,1,2,* Henrik Ståhl,3 Nicholas A. Kamenos,4 Gavin Turner,3 Heidi L. Burdett,4 and

Ronnie N. Glud1,2,3,5

1 University of Southern Denmark, Nordic Centre for Earth Evolution (NordCEE), Odense, Denmark

2 Greenland Institute of Natural Resources, Greenland Climate Research Centre, Nuuk, Greenland

3 Scottish Association for Marine Sciences, Scottish Marine Institute, Oban, United Kingdom

4 School of Geographical and Earth Sciences, University of Glasgow, Glasgow, United Kingdom

5 University of Århus, Arctic Research Centre, Århus, Denmark

* Corresponding author: [email protected]

Running title: Benthic O2 exchange in maerl and sand

145

Paper III Benthic O2 exchange in maerl and sand

146

Paper III Benthic O2 exchange in maerl and sand

Acknowledgements

We thank Anni Glud for constructing the oxygen microelectrodes used in this study and Daniel F.

McGinnis for providing the SOHFEA software package we used to estimate the EC fluxes. This project was financed by the Commission for Scientific Research in Greenland (KVUG; GCRC6507), the UK Natural Environmental Research Council (NERC, NE/F018612/1, NE/F0122991/1,

NE/G006415/1), the European Research Council through an Advanced Grant (ERC-2010-

AdG20100224) and the Danish National Research Foundation (DNRF53).

147

Paper III Benthic O2 exchange in maerl and sand

148

Paper III Benthic O2 exchange in maerl and sand

Abstract

Maerl beds are slow-growing, structurally complex perennial habitats that support a very rich biodiversity, yet in contrast to seagrass beds or kelp forests they are surprisingly understudied. We present the first rate estimates of benthic gross primary production (GPP), community respiration

(R), and net ecosystem metabolism (NEM) in a pristine maerl bed in temperate Loch Sween in different seasons from a series of benthic O2 exchange measurements using the non-invasive eddy correlation (EC) technique. For comparison, EC measurements were also carried out at a nearby sandy benthic habitat. The EC measurements document a highly dynamic O2 exchange rate driven by light availability during daytime and flow velocity at the seabed during the night. A substantial

-2 -1 GPP was evident at both sites, and varied seasonally from 0.7 to 1.8 mmol O2 m h at the maerl

-2 -1 site and from 0.4 to 3.3 mmol O2 m h at the sand site. Annual rates of GPP were very similar at

-2 -1 both sites (~12 mol O2 m yr ). Despite a substantial benthic primary production both the maerl bed and the sand habitat were heterotrophic year round. Additional inputs of carbon of ~4.4 mol m-2 yr-1 at the maerl site and ~7.9 mol m-2 yr-1 at the sand site were required to sustain the benthic

-2 -1 O2 demand. The compensation irradiance (Ic) varied seasonally from 16-75 µmol m s at the maerl site and from 13-140 µmol m-2 s-1 at the sand site, and was typically ~1.8 times lower at the

2 maerl bed. Parallel deployment of 0.1 m benthic chambers during night time resolved O2 uptake

-2 -1 rates that varied by up to ~8 times between replicate chambers (from -0.4 to -3.0 mmol O2 m h ; n=4). By contrast, the EC flux footprint was typically ~15 m2 at the maerl site and ~30 m2 at the sand site. Despite extensive small scale variability, mean rates of O2 uptake as resolved in parallel by chambers and EC in both benthic habitats were within 20% of one another.

149

Paper III Benthic O2 exchange in maerl and sand

150

Paper III Benthic O2 exchange in maerl and sand

Introduction

Maerl beds are composed of dense aggregations of free-living red corraline algae (Rhodophyta,

Corralinales) that accumulate on photic seabed surfaces. Maerl beds are slow-growing, structurally and functionally complex perennial habitats that support a rich diversity and biomass of autotrophs as well as benthic macrofauna (BIOMAERL team 2003; Grall et al. 2006), and are important nursery areas for juvenile invertebrates and fish (Kamenos et al. 2004a, 2004b). Maerl beds occur worldwide and are among the most widespread of the benthic phototrophic ecosystems in terms of areal coverage (Foster 2001). While their importance for regional biodiversity is increasingly being recognized, maerl beds remain severely understudied when compared to other conspicuous benthic ecosystems such as seagrass beds or kelp forests. One aspect that has received little attention is the productivity of the maerl community and its role in coastal carbon (C) cycling. This is surprising considering that the few existing studies on the topic indicate that maerl bed communities are among the most productive benthic ecosystems, with areal rates comparable to those of seagrass beds (Martin et al. 2005, 2007). Evidently, a better understanding of maerl beds and their role in coastal C cycling is required.

The most widely used method for estimating benthic primary production and C turnover at the seabed is the benthic O2 exchange rate (Glud 2008). Community rates of benthic gross primary production (GPP), respiration (R), and net ecosystem metabolism (NEM) have traditionally been estimated using benthic O2 chambers. Chambers enclose an area of sediment and derive the sediment exchange rate from the change in O2 concentration in the incubated water phase over time. Typically, chamber incubations enclose a sediment area of ~0.1 m2 and incubations last only for a few hours; therefore scaling up the derived rates to daily estimates assumes that the

151

Paper III Benthic O2 exchange in maerl and sand

integrated spatial and temporal scales of the chamber incubations are representative of the benthic community O2 dynamics at the seabed. Application of benthic chambers in maerl beds is complicated by the presence of the maerl structure and other large benthic macroalgae and fauna, and the enclosure process excludes the natural hydrodynamics that the incubated organisms would otherwise experience. More recently, the aquatic O2 eddy-correlation (EC) method by Berg et al. (2003) was introduced that allows estimating the benthic O2 exchange rate non-invasively and over large (~10-100 m2) areas of the seabed. The EC method combines high-frequency

Eulerian measurements of the flow velocity and the O2 concentration taken at a single point located typically 10-30 cm above the seabed surface to estimate the vertical turbulent exchange of

O2 between the seabed and the overlying water. Since EC measurements are carried out away from the seabed surface, this method is not confined to soft sediments. Importantly, EC measurements can provide continuous time series of O2 exchange rates covering one or more days, to help elucidate the dynamics of benthic O2 exchange under changing environmental conditions such as light. This method has been applied to complex coastal benthic habitats such as high-latitude rocky benthic surfaces (Glud et al. 2010), seagrass beds (Rheuban et al. 2014), oyster beds (Reidenbach et al. 2013), and tropical coral reefs (Long et al. 2013), and could therefore be a valuable tool for estimating the benthic O2 exchange rate in maerl beds.

The main objectives of this study were 1) to assess the productivity and C turnover in a pristine temperate maerl bed in different seasons using the EC method, 2) to compare the benthic O2 dynamics of the maerl bed with a neighbouring (within 20 m) sandy site exposed to the same environmental conditions (light, depth, flow velocity, temperature, salinity), and 3) to resolve the benthic O2 exchange rate using the EC method and benthic chambers in both environments.

152

Paper III Benthic O2 exchange in maerl and sand

Materials and methods

Study sites

The sites selected for this study were located in the tidal narrows of Caol Scotnish, a sheltered

~0.5 km2 side-loch within the Loch Sween Marine Protected Area on the west coast of Scotland

(Fig. 1). The shoreline rapidly descended to a depth of ~5 m at mean low water and this depth remained consistent across the narrows. Closest to the shoreline the seabed was characterized by bare sediments composed primarily of a mixture of sands, gravels, and relict shell fragments (Fig.

1). Here, the dominant macrofauna comprised the deposit-feeding black brittle star Ophiocomina nigra that occurred in densities of ~20 ind. m-2. The permeability of the top ~10 cm of sediments measured on intact cores as described by Klute and Dirksen (1986) was 1.7 x 10-11 ± 0.2 m-2 (mean

± SD, n=4). Further out, in the middle of the narrows, a pristine maerl bed of ~500 m length and

~40 m in width was present, characterised by a dense population of live maerl thalli

(Lithothamnion glaciale) and high densities (up to several hundred ind. m-2) of the suspension- feeding common brittle star Ophiothrix fragilis and the deposit-feeder Ophiocomina nigra. The top

5-10 cm of the maerl bed consisted of a highly porous, branched structure of live maerl thalli. This layer was underlain by a mixture of relict maerl skeletons and soft sediments that accumulated due to entrapment of finer particles by the maerl structures. The combined permeability of the maerl thalli and underlying ~10 cm of sediments measured on intact cores as described by Klute and Dirksen (1986) was 2.2 x 10-12 ± 0.3 m-2 (mean ± SD, n=4). The study sites were located within

~20 m of one another. The flow regime in Caol Scotnish was tidally-driven. Velocity magnitudes were most often within the range of 2-5 cm s-1. However, short (~2 hr) periods of intensified flow reaching 15-20 cm s-1 were also apparent and coincided with the incoming (flood) tide. Surface

153

Paper III Benthic O2 exchange in maerl and sand

wave heights were < 10 cm during our measurement campaigns. Macroalgae such as the whip-like

Chorda filum appeared seasonally in Caol Scotnish, with the highest densities observed during spring and summer. Deposits of degrading detritus gathered on the seabed in autumn.

In situ instrumentation

Eddy-correlation (EC) measurements

For each of the 5 measurement campaigns at Caol Scotnish we aimed to obtain a time-series of EC measurements at each of the two sites covering at least one complete 24 h period. During the

September 2010 and April 2012 campaigns two EC systems were available. This allowed for simultaneous measurements of EC O2 exchange rates at the two sites. During the other campaigns

(November 2011, February 2012, August 2012) a single EC system was alternated between the two sites. In all cases, EC deployment was done by divers that used lift bags to carefully position the EC systems on the seabed. A total of 14 successful EC deployments were made at Caol

Scotnish during the 5 measurement campaigns that altogether integrate >300 h of measuring time divided between the two sampling sites.

The configuration of the applied EC systems was similar to the original design by Berg et al (2003).

The main components of the EC systems consisted of an acoustic Doppler velocimeter (ADV,

Vector, Nortek) and a Clark-type O2 microsensor (10-20 µm tip diameter, a 90% response time ≤

0.5 s and a stirring sensitivity <0.5%) (Revsbech 1989, Gundersen et al. 1998) that relayed the signal to the ADV via a custom build submersible amplifier (McGinnis et al. 2011). The ADV recorded the streamwise (u), traverse (v), and vertical (w) flow velocity components as well as the

O2 microsensor data at frequencies of 32 or 64 Hz and in addition collected ancillary information such as instrument pitch and roll, flow direction, and signal strength. The equipment was mounted

154

Paper III Benthic O2 exchange in maerl and sand

onto a stainless-steel tripod frame measuring 130 cm by 90 cm, designed to minimize hydrodynamic interference. The ADV sampling volume was located ~15 cm above the seabed surface. Visual inspection by divers ensured that the O2 sensor tips and the ADV measurement volume stood well clear of bottom features such as stones, maerl branches, and benthic macrofauna. The O2 microsensor tip was carefully positioned just at the edge of the ADV sampling volume to extract the O2 data close to the ADV measurement point without compromising the velocity measurements. Prior to deployment, the EC O2 microsensors were left to polarize for ~12 hours to minimize sensor drift during deployments. Following polarization, the signal range and response time of each O2 microsensor used was evaluated using water samples of known O2 concentration. A sodium dithionite solution was used for the zero O2 saturation value, and collected bottom water samples were used for the in situ O2 concentration. The O2 concentrations were determined in the laboratory by Winkler titration.

Benthic chamber measurements

The square shaped benthic chambers constructed from Perspex and polyvinyl chloride enclosed an area of 961 cm2 (0.1 m2). The chambers had a removable lid that was fitted with a rotating cross- shaped stirrer on the underside. The stirrer was powered by a small self-contained battery- powered motor and was set to rotate at a constant frequency of ~17 revolutions per minute

(rpm). Several chambers were deployed by divers during the night in parallel at each of the two sites during periods that coincided with the EC deployments. The chambers were gently pushed

~15cm into the sediment and the chamber lid was sealed to prevent exchange with the ambient bottom water. A sampling port on the side of the chamber wall allowed water samples from inside the chambers to be retrieved during incubation. A 100 ml water sample was retrieved by the

155

Paper III Benthic O2 exchange in maerl and sand

divers at the start and at the end of each of the 3-6 h long incubations using a gas-tight syringe.

Once extracted, water samples were transported to shore where they were fixed, stored in the dark and later analyzed in triplicates for the O2 concentration by Winkler titration. The decline in

O2 saturation between the start and the end of the incubation was < 30%. Four chambers were deployed at each site in September, February and April. Three were deployed in November, and five in August. Benthic chambers were also deployed during the daytime, however from later analysis it was deduced that up to 80% of the ambient light was being attenuated by the chamber walls. Because of this, the chamber data that were collected during the daytime were omitted from further analyses, and instead only those collected during the night are presented.

In situ background environmental parameters

A small conductivity temperature depth (CTD) sensor (XR-420, RBR) equipped with an O2 optode

(3830, Aanderaa) and a scalar photosynthetically available radiation (PAR) sensor (QSP-2200,

Biospherical Instruments) was mounted onto a small stainless-steel frame that held the sensors 15 to 20 cm above the seabed. The CTD system was deployed by divers at the interface between the maerl site and the sandy site, which was approximately equidistant between the two measurement sites. The CTD was programmed to record the environmental parameters covering the benthic chamber and EC deployments at 30 s intervals.

Eddy–correlation fluxes

The EC O2 fluxes were first extracted from the raw EC dataset and then evaluated for their ‘quality’ based on a set of defined criteria as detailed in Attard et al. (2014). Flux extraction was carried out using the open source software package Sulfide Oxygen Heat Flux Eddy Analysis (SOHFEA) version

2.0 (available at www.dfmcginnis.com/SOHFEA). Additional data treatments that are detailed

156

Paper III Benthic O2 exchange in maerl and sand

below but not available in SOHFEA were carried out in MATLAB® (MathWorks). The datasets were processed for flux extraction in the following order: 1) Weak signals in the raw 32 or 64 Hz ADV velocity data were identified by a low beam correlation and/or low signal strength. Individual data points with beam correlations below 50% and signal-to-noise ratios below 12 dB were discarded.

2) The 32 or 64 Hz raw EC data was averaged down to 8 Hz. This reduced the noise level and the smaller file size allowed for easier data handling. 3) A spectral analysis was carried out on the 8 Hz vertical velocity ( ) and the O2 microsensor data ( ). The spectra showed the presence of an inertial subrange, identified as the region on the pressure spectral density plot for and where the slope of energy cascade to the smaller scales followed the predicted -5/3 fit, suggesting well- developed turbulence (Fig. 2). Inspection of the spectra and co-spectra for selected periods of high and low flow, as well as for periods with and without surface waves (<10 cm) suggested that there were no discernable abnormalities within the signals. The spectra were furthermore used to infer that the data reduction through adjacent averaging from 64 or 32 Hz to 8 Hz did not result in a loss of signal at high frequency, since most of the turbulent contributions typically occurred at a frequency of 2 Hz or lower. 4) The O2 microsensors were calibrated to the CTD O2 optode data. 5)

Spike noise in the velocity and O2 microsensor data were removed using the 3D phase space method by Mori et al. (2007). 6) The measured 8 Hz ADV velocity dataset was rotated using the

‘planar rotation’ method to obtain a vertical velocity component that is normal to the local

-2 -1 streamline (Lorke et al. 2013). 7) The O2 fluxes, in mmol m h , were extracted from the ADV velocity and O2 microsensor data as the covariance 〈 〉, where and are deviations from a least-squares linear trend fitted to the measured vertical velocity and O2 concentration respectively, and the angle bracket denotes time averaging. Following Reynolds decomposition theory the vertical velocity vector may therefore be expressed as 〈 〉 and the scalar

157

Paper III Benthic O2 exchange in maerl and sand

quantity 〈 〉 (Berg et al. 2003). The selected time averaging interval is a tradeoff between including as many of the flux-contributing turbulent eddies as possible while excluding low-frequency non-turbulent contributions such as advective flows that may compromise covariance statistics (McGinnis et al. 2008). To determine the optimal time interval, an analysis was carried out to investigate the effects of averaging time on the mean covariance and subsequent flux estimates for the friction velocity ( ) and the O2 fluxes. The is directly related to the turbulence regime within the BBL. Estimates for were computed from complex Reynolds stress measurements derived from the ADV velocity time series (McPhee 2008). The streamwise

(u), traverse (v), and vertical (w) velocity components were decomposed into a mean and deviatory velocity as 〈 〉 , 〈 〉 and 〈 〉 . The was then calculated

as 〈 〉 〈 〉 . The mean and O2 flux were computed as a function of the ensemble average. A time window of 10 min was consistently identified as the optimal interval for flux calculation at the two sites (Fig. 2). 8) A time-shift correction was applied to the data. Time shifting was performed for each ensemble interval by shifting the O2 data in time relative to the velocity data to a maximum of 2 seconds to achieve the maximum correlation (defined in terms of the maximum flux) for 〈 〉. This correction is necessary when the physical separation between the O2 sensor and the ADV measurement volume, and/or the sensor response time, result in a slight misalignment in the data (McGinnis et al. 2008). 9) By assuming law-of-the-wall velocity

profiles, the mean sediment surface roughness ( ) was estimated as , where

is the measurement height above the benthic surface (0.15 m), ҡ is the von Karman constant

(0.41), and is the flow velocity magnitude.

158

Paper III Benthic O2 exchange in maerl and sand

The extracted 10 min EC fluxes were then carefully evaluated for their quality based on three criteria, namely: 1) collisions of particles or debris with the O2 microsensor (spikes), 2) rapid changes in flow direction and/or flow velocity magnitude, and 3) suppressed exchange rates of O2 due to insufficient turbulent mixing, which typically occurred when dropped below ~2cm s-1.

Altogether the screening process based on the three criteria typically resulted in the exclusion of

<15 % of the measured 10 min EC O2 fluxes.

Dark benthic chamber O2 fluxes

-2 -1 Areal rates of benthic O2 uptake in mmol m h were derived from the rate of change of O2 in the well-mixed incubated water phase over time as (v/a)*(∂C/∂t), where v is the volume of the water

3 2 phase in the chamber (in m ), a is the area of enclosed sediment (in m ), C is the O2 concentration

(in mmol m-3), and t is the incubation time (in h) (Glud 2008).

Rates of benthic productivity

The rates of benthic productivity were calculated from each time series of the EC O2 exchange rates. Nighttime periods were identified as the periods when PAR was ≤ 1 µmol m-2 s-1. Daytime periods comprised the remaining intervals. The O2 exchange during light (termed ‘net daytime

-2 -1 production’, NDP) and R were then computed from the EC time series in mmol O2 m h as a bulk average of the O2 fluxes during light and dark, respectively. Assuming a light-independent

-2 -1 respiration rate we also derived estimates for the benthic GPP (in mmol O2 m hr ) as

| |. Acknowledging that the R rate during daytime typically is higher than that at night, we regard the GPP estimates presented herein as minimum values (Glud et al. 2009).

159

Paper III Benthic O2 exchange in maerl and sand

The autotrophic-heterotrophic balance of the benthic ecosystem (termed ‘net ecosystem metabolism’, NEM, in mmol m-2 d-1) was derived directly from the EC time series as a weighted average of the NDP and R fluxes accounting for the number of day versus night hours. The NEM indicated whether sediment O2 production through photosynthesis balanced the various heterotrophic processes that directly or indirectly consume O2. Positive NEM values indicated a net O2 release by the benthic ecosystem (autotrophy) while negative NEM values indicated a net

O2 uptake (heterotrophy) over a 24 h period.

Benthic community light response

The light dependency of the EC O2 fluxes was evaluated using the photosynthesis vs. irradiance (P-

I) relationship. The screened 10 min EC O2 fluxes covering at least one 24 h period were binned to

2 h intervals and plotted against the PAR data. This time interval was selected because it provided the best compromise between maintaining a high enough temporal resolution while minimizing the effects of e.g., non-steady state dynamics within the BBL that can substantially affect the EC O2 fluxes on shorter timescales (Holtappels et al. 2013). The relationship between the EC O2 fluxes and the PAR was investigated using various fitting functions and from regression analyses it was established that the P-I relationships were best explained by linear fits to the data. The

-2 -1 compensation irradiance ( , in μmol quanta m s ) was subsequently derived from each linear fit as the x-intercept.

160

Paper III Benthic O2 exchange in maerl and sand

Results

Metadata

Throughout the study the bottom water temperature ranged from 16.6 ± 0.4 ºC in the summer to

6.3 ± 0.2 ºC in the winter. Bottom water salinity varied between 26.4 ± 0.5 and 33.6 ± 0.2 and showed no seasonal trend. Bottom water O2 saturation was lowest in November (86.4 ± 2.4 %) and highest in April (107.1 ± 5.5 %) (Table 1). Variations in the bottom water O2 concentrations of

-1 up to 40 µmol O2 L (~15 % saturation) coincided with the tidal advection of O2-enriched waters from outer Loch Sween into Caol Scotnish. The daily integrated in situ PAR was lowest during the

February campaign (0.70 mol quanta m-2 d-1) and highest in April (5.48 mol quanta m-2 d-1).

Eddy-correlation measurements

The narrow (~100 m) width of Caol Scotnish and the tidally-driven flow resulted in well- constrained hydrodynamics that allowed the divers to accurately align the EC instrument’s streamwise (u) velocity component within the main flow direction. Therefore rotation of the traverse (v) velocity component during data processing was minimal and typically amounted to <

±10º. Care was taken to deploy the EC instruments level with the seabed such that the ADV tilt as measured by the instrument’s internal compass was < ±3º. When the flow velocity decreased below ~2cm s-1 the fluxes were seen to be highly suppressed, indicating that turbulent transport is also suppressed, and the eddy assumptions are no longer valid (Brand et al. 2008). Fluxes that fell within these periods were excluded from further analyses. However data exclusion due to insufficient turbulence amounted to only ~10% of the total time series since a residual flow with magnitudes around 2-5 cm s-1 was apparent in the transitions from flood to ebb tide and this generally maintained a steady flux signal. Linear relationships between the mean flow velocity

161

Paper III Benthic O2 exchange in maerl and sand

magnitude and indicate that a well-developed turbulent BBL was present during ~90% of the sampling time at the two sites (Fig. 3).

The EC O2 fluxes showed distinct responses to the changing environmental conditions at the seabed. Parallel deployment of two EC systems located ~20 m apart from one another documented substantial differences in the EC resolved O2 fluxes during both day and night between the two sites under very similar PAR, flow velocity magnitude, and bottom water O2 concentrations (Figs. 4, 5). Typically, during peak irradiance the EC O2 fluxes were positive, indicating an autotrophic benthic ecosystem. During the night the flow velocity magnitude exerted a major control on the EC fluxes, enhancing the rate of O2 uptake by ~10 times in the sandy site and ~4 times in the maerl site (Figs. 5, 6).

A substantial benthic primary production was detected at both sites year round. Mean rates of

-2 -1 -2 -1 GPP at the maerl site ranged from 0.65 mmol O2 m h (15.6 mmol O2 m d ) in November 2011

-2 -1 -2 -1 to 1.78 mmol O2 m h (42.7 mmol O2 m d ) in April 2012 (Table 2). At the sand site the GPP

-2 -1 -2 -1 -2 -1 -2 ranged from 0.37 mmol O2 m h (8.9 mmol O2 m d ) to 3.27 mmol O2 m h (78.5 mmol O2 m d-1). Due to sensor breakage ~6 h after deployment, no EC data were available at the sand site in

August during dark. Instead, O2 uptake rates from the chambers were used for the GPP calculation.

There was a net release of O2 from the seabed during the daytime at the maerl site in February,

-2 -1 -2 -1 April, and August 2012, up to 0.58 mmol O2 m h (13.9 mmol O2 m d ). In September 2010 and

November 2011 the maerl site was heterotrophic during the daytime. The mean EC O2 exchange

-2 -1 -2 -1 during dark at the maerl site ranged from -0.58 mmol O2 m h (-13.9 mmol O2 m d ) in

-2 -1 -2 -1 February 2012 to -2.16 mmol O2 m h (-51.8 mmol O2 m d ) in September 2010. By contrast,

162

Paper III Benthic O2 exchange in maerl and sand

the sand site was heterotrophic or close to metabolic balance during light year round (range from

-2 -1 -2 -1 0.04 to -0.98 mmol O2 m h , or 1.0 to -23.5 mmol O2 m d ). The mean EC O2 exchange during

-2 -1 -2 -1 the night at the sand site ranged from -0.48 mmol O2 m h (-11.5 mmol O2 m d ) in February

-2 -1 -2 -1 2012 to -3.69 mmol O2 m h (-88.6 mmol O2 m d ) in April 2012. When integrated over 24 h both the maerl site and the sand site were heterotrophic year round. At the maerl site the NEM

-2 -1 - ranged from -3.1 to -33.1 mmol O2 m d and at the sand site it was from -3.1 to -41.9 mmol O2 m

2 d-1.

The 2 h EC O2 fluxes showed a good correlation with the in situ PAR during all seasons (Fig. 7, Table

3). Linear relationships between the EC fluxes and the PAR suggest light under saturation of the benthic communities under the variety of field conditions we investigated. At the maerl site the slope of the linear P-I relation was consistently ~0.05 in the September 2010, November 2011, and

February 2012 datasets, and reduced to ~0.02 during April and August 2012. At the sand site the slope was more variable and ranged from ~0.01 in September 2010 to ~0.04 in February 2012. The

-2 -1 Ic calculated from the P-I relations as the x-intercept ranged from 16 to 75 µmol quanta m s at

-2 -1 the maerl site and from 13 to 140 µmol quanta m s at the sand site. At both sites, the highest Ic was observed during September 2010 and the lowest Ic was observed during February 2012. The Ic at the sand site was typically ~1.8 times higher than that at the maerl site, with the exception of

February 2012 where the Ic at the two sites were comparable.

Hydrodynamics and EC footprint characteristics

The and estimates computed from the ADV velocity time series were variable between sites and between seasons. However, the measurements consistently indicated a more intensified turbulent mixing at the maerl site compared to the sand site (Fig. 3). The mean for the 5

163

Paper III Benthic O2 exchange in maerl and sand

campaigns at the sand site was 0.33 cm ± 0.55 (mean ± SD, n=5) and at the maerl site it was 1.20 cm ± 1.23 (mean ± SD, n=5). The characteristics of the EC flux footprint for the sand (maerl) site calculated according to the parameterisation by Berg et al. (2007) suggest that the footprint was typically ~30 m (~15 m) long with a width of ~1 m and a region of maximum flux located ~1.3 m

(~0.5 m) upstream from the instrument. Due to the tidal reversal and therefore the reversal of the upstream location of the EC footprint, the EC O2 exchange estimates for GPP, R, and NEM actually integrate an area of about twice the size of the footprint dimensions estimated above. The spatial extent of the maerl bed and the sand site was on the order of hundreds of metres upstream of the

EC instrument in either direction of the flow, and therefore we can be confident that the flux estimates presented herein integrate the activities of each respective benthic ecosystem.

Benthic chamber measurements

The benthic O2 exchange rates from the chamber deployments during the night resolved O2 uptake rates at the maerl site ranging from -1.0 ± 0.1 mmol m-2 h-1 (mean ± SE, n=3) in November to -2.4 ± 0.2 mmol m-2 h-1 (mean ± SE, n=4) in September (Table 2). At the sand site the range was

-2 -1 -2 -1 from -0.5 ± 0.4 mmol O2 m h (mean ± SE, n=4) in February to -1.1 ± 0.2 mmol O2 m h (mean ±

SE, n=4) in April. A substantial variation in the resolved O2 exchange rate between replicate chambers was evident. On average, replicates varied by a factor of ~3 at the maerl site and by a factor of ~5 at the sand site. The highest variation between replicate chambers was observed at the sand site in April. Here, chamber O2 exchange rates varied by a factor of ~8, from -0.4 to -3.0

-2 -1 mmol m h . These differences are likely due to the spatial heterogeneity of O2-consuming processes at the seabed caused by uneven distribution of organisms on and within the sediment or by different sediment permeability- see Discussion.

164

Paper III Benthic O2 exchange in maerl and sand

Discussion

Benthic metabolism in a pristine live maerl bed

Detailed studies on the metabolism of isolated red corraline algae indicate low rates of photosynthesis and respiration due to their slow growth rate of ~1 mm yr-1 (Martin et al. 2013 and references therein; Blake & Maggs 2003). However, pristine maerl beds increase the substratum heterogeneity and provide structurally complex perennial habitats that are colonized by a large diversity and biomass of both autotrophic and heterotrophic organisms, substantially increasing the benthic production of the maerl bed (BIOMAERL team 2003). The complex structure and the spatial variability within a maerl bed makes estimating GPP, R, and NEM using benthic chambers challenging, and in this regard the EC method holds clear advantages. This study is the first to apply EC principles to estimate the benthic GPP, R, and NEM in a pristine maerl bed, and provides new insights into the benthic O2 dynamics at these sites that are valued and widespread features of coastal benthic ecosystems.

The non-invasive EC measurements documented a substantial benthic primary production in the

-2 -1 -2 -1 maerl bed year round. The highest mean rates of GPP of 1.78 mmol O2 m h (43 mmol O2 m d ) were measured in April and corresponded with the high levels of in situ PAR measured during this

-2 -1 -2 - campaign of ~5.5 mol quanta m d . The mean GPP was lowest in November, 0.65 mmol O2 m h

1 -2 -1 -2 -1 (16 mmol O2 m d ), when the PAR was ~4.7 fold lower (1.2 mol quanta m d ). Seasonal measurements in a temperate maerl bed in the Bay of Brest, France, using in situ benthic

-2 -1 chambers reported GPP values ranging from ~4 to 134 mmol O2 m d between winter and

-2 -1 summer and an annual rate of 24.5 mol O2 m yr (Martin et al. 2007). This is ~2.1 times higher

-2 -1 than the yearly GPP estimate for Caol Scotnish of 11.6 mol O2 m yr . However, the PAR available

165

Paper III Benthic O2 exchange in maerl and sand

for benthic photosynthesis in Loch Sween was up to ~8 times lower than that reported by Martin et al. (2007) for the Bay of Brest; this discrepancy could explain much of the observed variation in the GPP between the two temperate sites. Similarly, the seasonal range of daily R rates for the

-2 -1 maerl bed in Loch Sween (-14 to -53 mmol O2 m d ) are at the lower end of those reported from benthic chamber measurements in the same study by Martin et al. (2007) (from -33 to -163 mmol

-2 -1 O2 m d ) although comparable to the global mean sediment O2 uptake rate for marine sediments

-2 -1 at ~5 m depth (-31 mmol O2 m d ; Middelburg et al. 2005).

Despite a substantial benthic primary production, the maerl bed community was heterotrophic year round and required an additional input of organic matter that amounted to ~4.4 mol C m-2 yr-

1 (assuming a respiration quotient of 1.0). Depth-integrated annual rates of pelagic production in a nearby loch measured using 14C-labelled bicarbonate incubations found rates of up to ~17 mol C m-2 yr-1 (Rees et al. 1995). Therefore the annual C deficit at the seabed is likely to be maintained through sedimentation and entrapment of phytoplankton and to a lesser extent terrestrial derived material from the catchment area of Loch Sween. These inferences are consistent with the seasonal study by Martin et al. (2007) in the Bay of Brest, whereby despite a substantial annual benthic GPP the maerl community was heterotrophic and required an additional input of C amounting to 13.9 mol m-2 yr-1 (166.5 g C m-2 yr-1). At this location, Grall et al. (2006) investigated the origin of the carbon sources consumed by the main benthic macro- and mega-faunal species in the maerl bed by coupling community structure analyses and carbon and nitrogen isotopic composition. The study identified soft macroalgae, phytoplankton, and microphytobenthos as the three main carbon sources for primary consumers, highlighting the importance of sedimentation and entrapment of organic matter in maintaining the C deficit in the maerl bed.

166

Paper III Benthic O2 exchange in maerl and sand

In Caol Scotnish, the importance of external inputs of organic matter to the maerl bed is further substantiated by the large densities of the suspension-feeding brittle stars Ophiothrix fragilis associated with the live maerl. The brittle stars were seen to anchor themselves to the maerl thalli and raise their feeding arms into the turbulent water column (Fig. 1), thereby increasing the particle encounter rate (Allen 1998). Studies on dense beds (up to 2000 ind. m-2) of the suspension feeding brittle star Ophiothrix fragilis in the north east Atlantic indicate that optimal suspension- feeding activity is expressed when the flow velocity at the seabed is < 20 cm s-1, above which distortion of the feeding arm array occurs (Davoult & Gounin 1995). The flow velocity in Caol

Scotnish as measured by the EC system 15 cm above the maerl bed typically ranged from 2-15 cm s-1, and therefore the heterogeneous substrate of live maerl thalli combined with the hydrodynamics at the seabed in Caol Scotnish make this habitat ideal for the suspension-feeding brittle stars.

Comparison between maerl and sandy benthic habitats

Model studies indicate that the benthic NEM, the difference between benthic GPP and R, is positive in over 33% of the global shelf area (Gattuso et al. 2006). Although this may be the case, the global database on NEM, GPP, and R remains limited, particularly for seabed surfaces other than muddy sediments. Complex light-exposed benthic ecosystems such as maerl beds and permeable sediments cover large areas of the inner shelf region and may therefore substantially contribute to local and regional carbon cycling.

The tidal narrows of Caol Scotnish are suitable for investigating the benthic O2 dynamics in maerl and sandy habitats using EC. The two habitats were located at the same depth and within ~20 m of one another, and therefore experienced the same environmental conditions of e.g., PAR, flow

167

Paper III Benthic O2 exchange in maerl and sand

velocity, O2 concentrations, and temperature, and allowed for direct comparison under changing environmental conditions. The availability of PAR at the seabed is the primary requirement for benthic photosynthesis since the efficient mineralization of organic material at the seabed usually maintains a steady supply of nutrients that may be utilized by the benthic algae (Barranguet et al.

1998). The linear P-I relationships (Fig. 7; Table 3) document that the benthic communities at both the maerl and the sand site were light limited and probably did not experience a net light saturation even when in situ PAR exceeded 200 µmol quanta m-2 s-1. Red corraline algae inhabit a great diversity of environments ranging from shallow tropical reefs to the deepest reaches of the photic zone, and are typically low-light adapted (Littler et al. 1991; Roberts et al. 2002; Chisholm

2003). Assessment of the photosynthetic performance of L. glaciale in Loch Sween carried out in situ by pulse amplitude modulation (PAM) fluorometry suggest a low light adaptation of the

-2 -1 organisms with light saturation parameters (Ik) ranging from 4.5 to 54.6 µmol quanta m s

(Burdett et al. 2012). However, the EC measurements integrate also the contributions from the associated fauna and algae, and therefore the P-I relations represent the net community light response, including the advective flow of interfacial solutes within the surface sediments and the entrapment of organic carbon. At the maerl site the slope of the P-I relation was consistently

~0.05 during the 3 campaigns with lowest light (September 2010, November 2011, and February

2012) and then during the 2 other campaigns (April and August 2012), when the integrated daily

PAR was up to ~8 times higher, the slope was reduced to ~0.02. At the sand site the slope of the P-

I relations was more variable although the highest value of ~0.04 was observed during February

2012, when light was at its lowest. A seasonal variability in the compensation irradiance (Ic), that is, the PAR level at which GPP balances R was also observed (Table 3). At both sites the Ic was lowest during the winter as a result of the reduced R rates (February 2012; Tables 2 and 3). During

168

Paper III Benthic O2 exchange in maerl and sand

the remaining campaigns the Ic was substantially higher. The Ic at the sand site was consistently

~1.8 times higher than that at the maerl bed and therefore lower levels of in situ PAR were required to stimulate a net autotrophic response at the maerl site compared to the sand site (Fig.

7, Table 3).

Temperature and irradiance exert major controls on GPP and R in benthic environments, and subsequently also on the derived P-I parameters (Hancke et al. 2014). Seasonal changes in water temperature and PAR reaching the seabed may explain much of the variability of GPP and R in some benthic settings (Martin et al. 2007). However, the seasonal response is frequently confounded by other variables such as changes in the benthic community structure and biomass

(Barranguet et al. 1998, Attard et al. 2014). While maerl is slow-growing and therefore changes in its biomass over seasons are assumed negligible (Martin et al. 2005), the biomass of associated fauna and algae may change substantially (Grall et al. 2006). For example, transient accumulations of macroalgae were evident in Loch Sween, and this likely altered the NEM as resolved directly by the EC measurements during both the periods of macroalgal growth as well as later on in the season when the degrading detrital material was deposited on the seabed (Dalsgaard 2003).

Overall, the seasonal variability in benthic GPP, R, and NEM underlines the importance of considering seasonal measurements when resolving the annual benthic ecosystem metabolism in shallow coastal settings.

Benthic chamber versus EC measurements

The benthic ecosystems in Loch Sween, comprised principally of permeable sediments and a patchy distribution of large epifauna and macroalgae, would suggest substantial spatial heterogeneity in the O2-consuming processes during dark. This is evidenced by the variations

169

Paper III Benthic O2 exchange in maerl and sand

between replicate 0.1 m2 chambers by up to a factor of ~8. Despite the large variability between individual chambers, the mean rates of O2 uptake during dark estimated from the chamber deployments were in most cases comparable to those that were resolved in parallel by EC (Table

4). In general, the benthic chamber O2 uptake rate estimates were higher than those measured by

EC at the maerl site, but lower at the sand site. At the maerl site, the ratio of the mean benthic O2 uptake rate as resolved by EC to the chambers ranged from 0.6 to 1.5 with an overall average of

1.0. At the sand site the range was from 0.9 to 3.4 with an overall average of 1.7. We note that the largest discrepancies between the EC and the chamber data were measured in April, when a substantial biomass of macroalgae was seen to have gathered on the seabed and especially on the sandy site. During this campaign, small areas of the sediment surface were cleared of the largest pieces of debris by the divers prior to installing the chambers in order to ensure a good seal between the chamber walls and the sediment. While properly sealing the chamber is essential to get reliable measurements of the O2 exchange, removal of the surficial detritus could explain much of the variations between the benthic O2 uptake rates as resolved by EC measurements (that include the detritus) and the chambers in April. Excluding the April datasets from the annual average EC:chamber rates would give a ratio of 0.8 ± 0.2 (mean ± SD, n=4) at the maerl site and 1.2

± 0.2 (mean ± SD, n=3) at the sand site for the remaining campaigns.

The application of chamber versus EC measurements has been investigated in the literature within the context of different benthic settings. Considerable differences between the EC and chamber resolved O2 exchange rates have typically been reported for so-called ‘complex’ benthic surfaces such as highly permeable sediments, where benthic chambers are not able to accurately recreate the natural pore-water flushing of the upper layers of the sediments that would occur in situ under high flow velocities (Berg et al. 2013). On the other hand, EC and chamber resolved O2

170

Paper III Benthic O2 exchange in maerl and sand

exchange rates are reported to be very similar for settings such as bare deep sea sediments that lack much of the complexity of their shallow water counterparts (Berg et al. 2009; Table 4). At the maerl site, the sediment permeability of 2.2 x 10-12 ± 0.3 m2 is at the interface between viscous- and advective-dominated O2 transport. At the sand site the sediments were more permeable (1.7 x 10-11 ± 0.2 m2) and this would allow a considerable advective flow of pore-waters within the sediments and exchange with the overlying water column (Huettel and Gust 1992). The O2 exchange rate for permeable sediments is dependent upon the hydrodynamic conditions, whereby an increase in flow velocity increases the oxic volume of the surface sediments temporarily enhances the rate of O2 uptake (Cook et al. 2007, Berg et al. 2013). Consistent with these observations, the rate of O2 uptake during dark increased with flow at both sites (Figs. 5, 6).

Furthermore, the relationship between the flow velocity magnitude and the dark EC O2 exchange rates indicate higher flow-induced stimulation of O2 uptake at the sand site compared with the maerl site that may at least in part be attributed to higher sediment permeability.

While the EC measurements integrate the natural hydrodynamics and are therefore ideally placed to investigate the O2 exchange rate in permeable sediments (Berg et al. 2013), the benthic chambers isolate the incubated sediment and associated organisms from the surrounding environment and the rotating stirrer defines a turbulence regime within the chamber that is markedly different from those in situ. The hydrodynamics may influence various components of the benthic ecosystem. For example, high flow velocity at the seabed can enhance rates of benthic primary production by (1) facilitating the efflux of O2 from the phototrophic organisms and thus increasing the affinity of RuBisCO to CO2, increasing gross primary production (Mass et al. 2010), and (2) by facilitating influx of nutrients and DIC required for photosynthesis (Enriquez & Rodriguez-Roman 2006; Cook & Røy

2006). For highly permeable sediments, a tight coupling exists between the benthic O2 uptake rate and the

171

Paper III Benthic O2 exchange in maerl and sand

flow velocity. Increased flow stimulates advective flushing of surface sediments with oxygenated water, increases the benthic O2 uptake rate (Cook et al. 2007; Berg et al. 2013), and promotes the decomposition of dissolved organic carbon (DOC) (Chipman et al. 2010). The hydrodynamics also influences the feeding behaviour of benthic fauna such as suspension-feeding brittle stars by increasing particle encounter rates through increased turbulence and resuspension of settled material (Davoult & Gounin 1995). In light of these differences, the fact that the chamber and EC budgets during dark are within ~20% of one another lends credibility to the application of both these methods in such environments. Similar findings have been reported for other shallow water studies that combined EC and chamber measurements (Table 4).

Overall, direct interpretation of short-term variations in EC O2 exchange rates as a true benthic O2

(C) flux during both light and dark periods is complicated by a non-steady state distribution of interstitial O2 and reduced inorganic products within the sediments under dynamic conditions

(Glud et al. 2007). Therefore the general approach has been to average the derived EC O2 exchange rates (in our case, 10 min) into longer time windows of 1 h or more to reduce the major short-term variations in the O2 exchange that are due to non-steady state within the sediment or the BBL, and to provide a better approximation of the C turnover over days and seasons as a function of environmental drivers such as PAR (Attard et al. 2014; Rheuban et al. 2014).

Furthermore, long measurement timescales are important to integrate transient activity of benthic fauna that can markedly alter the benthic O2 exchange (Wenzhöfer and Glud 2004), and that could otherwise easily be omitted when using e.g., chamber incubations that integrate just a few hours. As evidenced in the variability between chamber replicates, substantial spatial variability in the O2 consuming processes within and on the sediment on a metre scale documents the importance of replication. It is noteworthy that even in the absence of large macro fauna and

172

Paper III Benthic O2 exchange in maerl and sand

algae, chamber replicates often gave substantially different estimates of the O2 uptake, suggesting that spatial variations within rather than on the sediment were also important.

Implications for future studies

We have used diel time series of aquatic EC O2 exchange to infer the benthic GPP, R, and NEM in maerl and sand habitats in temperate Loch Sween in different seasons. We have documented that sufficient turbulent mixing was present at both sites to allow O2 exchange rates to be derived from

~90% of each 24 h EC time series. The measurements document a dynamic O2 exchange rate that on short timescales (minutes to hours) is driven mainly by the flow velocity magnitude during night and by PAR during day. This suggests that the environmental conditions at the time of sampling may have large effects on the resolved exchange rates, and therefore, long timescales need to be integrated to account for variability in e.g., weather conditions. Our measurements typically integrated 24-48 h of near-continuous EC measurements of O2 exchange per site for each campaign, and although these measurements currently represent some of the best estimates of

GPP, R, and NEM in such environments, longer term deployments will provide even better estimates. Integration of more robust and stable O2 microsensors such as microoptodes (Chipman et al. 2012) will be a key development in order to improve chances of success of long-term EC deployments. Furthermore, while the estimates for EC and chamber measurements were within

20% of one another, this comparison should be interpreted cautiously since both methods integrate different scales of space and time.

173

Paper III Benthic O2 exchange in maerl and sand

174

Paper III Benthic O2 exchange in maerl and sand

References

Allen, J. R. 1998. Suspension feeding in the brittle-star Ophiothrix fragilis: efficiency of particle retention and implications for the use of encounter-rate models. Marine Biology 132: 383-390.

Attard, K. M., R. N. Glud, D. F. McGinnis, and S. Rysgaard. 2014. Seasonal rates of benthic primary production in a Greenland fjord measured by aquatic eddy correlation. Limnol. Oceanogr. 59: 1555-1569.

Barranguet, C., J. Kromkamp, and J. Peene. 1998. Factors controlling primary production and photosynthetic characteristics of intertidal microphytobenthos. Mar. Ecol. Prog. Ser. 173: 117-126.

Berg, P., H. Roy, and P. L. Wiberg. 2007. Eddy correlation flux measurements: The sediment surface area that contributes to the flux. Limnol. Oceanogr. 52: 1672-1684.

Berg, P., and M. Huettel. 2008. Monitoring the seafloor using the noninvasive eddy correlation technique: Integrated benthic exchange dynamics. Oceanography 21: 164- 167.

Berg, P., R. N. Glud, A. Hume, H. Stahl, K. Oguri, V. Meyer, and H. Kitazato. 2009. Eddy correlation measurements of oxygen uptake in deep ocean sediments. Limnol. Oceanogr.: Methods 7: 576–584.

BIOMAERL team. 2003. Conservation and management of northeast Atlantic and Mediterranean maerl beds. Aquatic Conserv.: Mar. Freshw. Ecosyst. 13: S65-S76.

Blake, C., and C. A. Maggs. 2003. Comparative growth rates and internal banding periodicity of maerl species (Corralinales, Rhodophyta) from northern Europe. Phycologia 42: 606- 612.

175

Paper III Benthic O2 exchange in maerl and sand

Burdett, H. L., S. J. Hennige, F. T. -Y. Francis, and N. A. Kamenos. 2012. The photosynthetic characteristics of red corraline algae, determined using pulse amplitude modulation (PAM) fluorometry. Botanica Marina 55: 499-509.

Chisholm, J. R. M. 2003. Primary productivity of reef-building crustose corraline algae. Limnol. Oceanogr. 48: 1376-1387.

Dalsgaard, T. 2003. Benthic primary production and nutrient cycling in sediments with benthic microalgae and transient accumulation of macroalgae. Limnol. Oceanogr. 48: 2138-2150.

Foster, M. S. 2001. Rhodoliths: between rocks and soft places. J. Phycol. 37: 659-667.

Glud, R. N. 2008. Oxygen dynamics of marine sediments. Mar. Biol. Res. 4: 243–289.

Glud, R. N., J. Woelfel, U. Karsten, M. Kuhl, and S. Rysgaard. 2009. Benthic microalgal production in the Arctic: Applied methods and status of the current database. Bot. Mar. 52: 559-678.

176

Paper III Benthic O2 exchange in maerl and sand

Grall, J., F. Le Loc’h, B. Guyonnet, and P. Riera. 2006. Community structure and food web based on stable isotopes (δ15N and δ13C) analysis of a North Eastern Atlantic maerl bed. J. Exp. Mar. Biol. Ecol. 338: 1-15.

Gundersen, J. K., N. B. Ramsing, and R. N. Glud. 1998. Predicting the signal of O-2 microsensors from physical dimensions, temperature, salinity, and O-2 concentration. Limnol. Oceanogr. 43: 1932-1937.

Hancke, K., B. K. Sorell, L. C. Lund-Hansen, M. Larsen, T. Hancke, and R. N. Glud. 2014. Effects of temperature and irradiance on a benthic microalgal community: A combined two- dimensional oxygen and fluorescence imaging approach. Limnol. Oceanogr. 59: 1599– 1611.

Holtappels, M., R. N. Glud, D. Donis, B. Liu, A. Hume, F. Wenzhoefer, and M. Kuypers. 2013. Effect of transient bottom water currents and oxygen concentrations on benthic exchange rates as assessed by eddy correlation measurements. J. Geophys. Res.: Oceans 118: 1157-1169.

Huettel, M., and G. Gust. 1992. Impact of bioroughness on interfacial solute exchange in permeable sediments. Mar. Ecol. Prog. Ser. 89: 253–267.

Kamenos, N. A., P. G. Moore, and J. M. Hall-Spencer. 2004a. Nursery-area function of maerl grounds for juvenile queen scallops Aequipecten opercularis and other invertebrates. Mar. Ecol. Prog. Ser. 274: 183-189.

Kamenos, N. A., P. G. Moore, and J. M. Hall-Spencer. 2004b. Small-scale distribution of juvenile gadoids in shallow inshore waters; what role does maerl play? ICES J. Mar. Sci. 61: 422-429.

Littler, M. M., D. S. Littler, and M. D. Hanisak. 1991. Deep-water rhodolith distribution, productivity, and growth history at sites of formation and subsequent degradation. J. Exp. Mar. Biol. Ecol. 150: 163-182.

Klute, A., and C. Dirksen. 1986. Hydraulic conductivity and diffusivity: Laboratory methods, p. 687-734. In A. Klute [ed.], Methods of soil analysis-part 1-physical and mineralogical methods. American Society of Agronomy.

Long, M. H., P. Berg, D. de Beer, and J. C. Zieman. 2013. In situ coral reef oxygen metabolism: An eddy correlation study. PloS one 8: e58581, doi:10.1371/journal.pone.0058581

177

Paper III Benthic O2 exchange in maerl and sand

Martin, S., J. Clavier, J. M. Guarini, L. Chauvaud, C. Hily, J. Grall, G. Thouzeau, F. Jean, and J. Richard. 2005. Comparison of Zostera marina and maerl community metabolism. Aquatic Botany 83: 161-174.

Martin, S., J. Clavier, L. Chauvaud, and G. Thouzeau. 2007. Community metabolism in temperate maerl beds. I. Carbon and carbonate fluxes. Mar. Ecol. Prog. Ser. 335: 19- 29.

Martin, S., A. Charnoz, and J-P Gattuso. 2013. Photosynthesis, respiration and calcification in the Mediterranean crustose corraline alga Lithophyllum cabiochae (Corralinales, Rhodophyta). Eur. J. Phycol. 48: 163-172.

McGinnis, D. F., P. Berg, A. Brand, C. Lorrai, T. J. Edmonds, and A. Wüest. 2008. Measurements of eddy correlation oxygen fluxes in shallow freshwaters: Towards routine applications and analysis. Geophys. Res. Lett. 35: L04403, doi:10.1029/2007GL032747

McPhee, M. 2008. Air-ice-ocean interaction: turbulent ocean boundary layer exchange processes. Springer.

Middelburg, J. J., C. M. Duarte, and J. P. Gattuso. 2005. Respiration in coastal benthic communities, p. 206–225. In P. A. Del Giorgio and P. J. L. B. Williams [eds.], Respiration in aquatic ecosystems. Oxford University Press.

Mori, N., T. Suzuki and S. Kakuno. 2007. Noise of acoustic Doppler velocimeter data in bubbly flow. Journal of Engineering Mechanics 133: 122-125.

Rees, A. P., N. J. P. Owens, M. R. Heath, D. H. Plummer, and R. S. Bellerby. 1995. Seasonal nitrogen assimilation and carbon fixation in a fjordic sea loch. J. Plankton Res. 17: 1307-1324.

178

Paper III Benthic O2 exchange in maerl and sand

Revsbech N. P. 1989. An oxygen microelectrode with a guard cathode. Limnol. Oceanogr. 34: 474-478.

Rheuban, J. E., and P. Berg. 2013. The effects of spatial and temporal variability at the sediment surface on aquatic eddy correlation flux measurements. Limnol. Oceanogr.: Methods 11: 351-359.

Rheuban, J. E., P. Berg, and K. J. McGlathery. 2014. Multiple timescale processes drive ecosystem metabolism in eelgrass (Zostera marina) meadows. Mar. Ecol. Prog. Ser. 507: 1-13.

Roberts, R. D., M. Kühl, R. N. Glud, and S. Rysgaard. 2002. Primary production of crustose corraline red algae in a high Arctic fjord. J. Phycol. 38: 273-283.

Wenzhöfer, F., and R. N. Glud. 2004. Small-scale spatial and temporal variability in coastal

benthic O2 dynamics: Effects of fauna activity. Limnol. Oceanogr. 49: 1471-1481.

179

Paper III Benthic O2 exchange in maerl and sand

180

Paper III Benthic O2 exchange in maerl and sand

Tables

Table 1: Metadata for the CTD deployments for the five measurement campaigns. Values for depth, temperature, salinity, and O2 are presented as a bulk average of the entire deployment (±

SD). The SD therefore reflects the variability of each parameter throughout the deployment rather than measurement accuracy.

Measurement O PAR Depth (m) Temperature (°C) Salinity 2 O (% sat.) campaign (μmol L-1) 2 (mol m-2 d-1) Sep 2010 4.7 ± 0.3 14.6 ± 0.2 32.3 ± 0.3 244.5 ± 19.6 96.2 ± 7.6 0.95 Nov 2011 5.0 ± 0.4 10.1 ± 0.3 30.8 ± 0.5 244.2 ± 7.3 86.4 ± 2.4 1.17 Feb 2012 4.7 ± 0.4 6.3 ± 0.2 31.6 ± 0.2 309.6 ± 6.4 101.2 ± 1.7 0.70 Apr 2012 4.6 ± 0.3 9.6 ± 0.4 33.6 ± 0.2 302.2 ± 10.6 107.1 ± 5.5 5.48 Aug 2012 4.7 ± 0.3 16.6 ± 0.4 26.4 ± 0.5 270.3 ± 9.3 106.3 ± 3.6 3.29

181

Paper III Benthic O2 exchange in maerl and sand

Table 2: EC O2 exchange rates (mean ± SE). The values for light fluxes (NDP), dark fluxes (R), GPP,

and chambers are in mmol m-2 h-1 whereas NEM is in mmol m-2 d-1. The GPP for the sand site in

August is estimated using the R value from the chamber incubations since no nighttime EC data

was available.

Measurement Maerl site Maerl Sand site Sand campaign NDP R GPP NEM Chambers NDP R GPP NEM Chambers Sep 2010 -0.69±0.4 -2.16±0.4 1.49 -33.1 -2.36±0.19 0.04±0.3 -1.09±0.5 1.13 -11.7 -0.90±0.05 Nov 2011 -0.30±0.3 -0.95±0.2 0.65 -15.5 -1.05±0.14 -0.98±0.2 -1.42±0.5 0.44 -29.2 -1.07±0.28 Feb 2012 0.58±0.3 -0.58±0.1 1.16 -4.1 -1.05±0.38 -0.03±0.1 -0.48±0.2 0.45 -7.8 -0.51±0.21 Apr 2012 0.15±0.3 -1.63±0.3 1.78 -13.7 -1.06±0.19 -0.42±1.5 -3.69±1.3 3.27 -41.9 -1.08±0.64 Aug 2012 0.47±0.2 -1.04±0.2 1.51 -3.1 -1.11±0.15 -0.08±0.6 - 0.37* -3.1 -0.45±0.12

182

Paper III Benthic O2 exchange in maerl and sand

Table 3: The fitting statistics of the linear P-I relationships as fitted to 2 h averages of the EC O2 exchange rates. No P-I relationship is available for the sand site in August 2012 due to sensor damage incurred after just ~6 h of deployment.

Measurement Maerl site Sand site campaign 2 2 Slope Intercept Ic R Slope Intercept Ic R Sep 2010 0.05 3.4 75 0.96 0.01 1.7 140 0.99 Nov 2011 0.05 2.2 47 0.84 0.02 1.7 85 0.43 Feb 2012 0.05 0.8 16 0.84 0.04 0.5 13 0.47 Apr 2012 0.02 0.9 61 0.84 0.02 2.6 109 0.62 Aug 2012 0.02 1.2 63 0.92 - - - -

183

Paper III Benthic O2 exchange in maerl and sand

Table 4: A summary of studies in various benthic settings that employed both chamber and EC measurements to estimate the benthic O2 uptake rate during dark. The flow velocity is measured by the EC instrument 10-15 cm above the seabed surface. The EC and chamber O2

-2 -1 uptake rates are presented in mmol m d . The EC O2 uptake values are reported as a bulk average of the O2 exchange rates over the length of the dataset. The values for Loch Sween, Scotland, are bulk averages from the seasonal measurements presented in this manuscript. The April 2012 measurements were excluded from the overall average due to the benthic chamber measurements being compromised during this campaign.

Depth Sediment EC dataset Flow velocity Chambers (no. of EC O Chamber O Study Location 2 2 EC:Ch (m) type length (h) (cm s-1) replicates, surface area) uptake uptake Aarhus Bay, Cohesive Berg et al. 2003 12 0.7 2 4, 0.1 40 28 1.4 Denmark mud Limfjord, Cohesive Berg et al. 2003 8 1 1.5-4.0 6, 0.1 45 40 1.1 Denmark mud Permeable Berg & Huettel 2008 Florida, USA 1 11 0-5 5, 0.1 368 98 3.8 sands Cohesive Berg et al. 2009 Sagami Bay, Japan 1450 1.5 1.0-3.4 2, 0.1 1.62 1.65 1.0 mud Permeable Reimers et al. 2012 Oregon, USA 80 31 0.3-13.4 6, 0.04 5.6 5.9 0.9 sands Wakulla River, Permeable Berg et al. 2013 3 2 31 5, 0.1 360 89 4.1 USA sands Tommeliten, Permeable McGinnis et al. 2014 74 40 1-10 1, 0.1 10.3 7 1.5 North Sea sands Loch Etive, Cohesive Glud et al. unpubl. 50 44 1-13 4, 0.1 12 14 0.9 Scotland mud Loch Sween, This study 5 Maerl bed 54 2-20 16, 0.1 28 33 0.8 Scotland Loch Sween, Permeable This study 5 34 2-20 10, 0.1 24 20 1.2 Scotland sands 184

Paper III Benthic O2 exchange in maerl and sand

185

Paper III Benthic O2 exchange in maerl and sand

Figure legends

Fig. 1: Map indicating the location of Caol Scotnish in Loch Sween on the west coast of Scotland

(A). The images show the two study sites investigated: (B) the sand/gravel site, and (C) the pristine maerl bed. Sources: Scottish Natural Heritage and Rob Cook.

Fig. 2: The analyses carried out to investigate the impact of averaging time on the mean estimate for (A) friction velocity ( ) and (B) O2 exchange for both sites. The data used to derive these estimates covers a 3 h period during dark when EC instruments were measuring in parallel at both sites. The mean (±SE) flow velocity magnitude ( ) during this period was 6.9 ± 0.5 cm s-1 at the maerl site and 8.5 ± 0.5 cm s-1 at the sand site. The same analysis was done for periods of lower flow velocity magnitude. Around 90% of the turbulent contributions occurred at frequencies between 1-0.01 Hz, or 1-100 s. Mean exchange rates were independent of averaging time after 10 min, and therefore this time interval was selected for flux extraction. The spectrum of (C) the vertical velocity component and (D) the O2 concentration indicate good agreement with the predicted -5/3 fit to the inertial sub range, suggesting well-developed turbulence.

Fig. 3: Friction velocity ( ) as a function of flow velocity magnitude ( ) from parallel EC deployments over a 24 h period in April. The linear relationship indicates a well-developed turbulent BBL. Despite very similar flow magnitudes at both sites, the was consistently higher at the maerl bed, indicating more vigorous turbulent mixing due to higher surface roughness.

Fig. 4: Eddy correlation O2 exchange rates during light at (A) the sand site and (B) the maerl site.

The measurements indicate a dynamic O2 exchange rate on timescales of minutes in response to

(E) changes in down welling PAR at the seabed.

186

Paper III Benthic O2 exchange in maerl and sand

Fig. 5: Eddy correlation O2 exchange rates during dark at (A) the sand site and (B) the maerl site.

On short timescales, the flow velocity magnitude was the major control on the of O2 exchange.

Fig. 6: The 10min EC O2 exchange rates presented in Fig. 5 as a function of (A, B) PAR during the daytime, and (C, D) the flow velocity magnitude during dark.

Fig. 7: P-I relationships for both sites over the year. Linear regressions were fitted to EC O2 exchange rates that were binned from the original 10 min intervals to 2 h or more to reduce the variability that was independent of PAR. With the exclusion of the February datasets, the compensation irradiance at the maerl site was consistently ~1.8 times lower than that at the sand site. No P-I relationship is available for the sand site in August due to sensor breakage after ~6 h of deployment.

187

Paper III Benthic O2 exchange in maerl and sand

Fig. 1

188

Paper III Benthic O2 exchange in maerl and sand

Fig. 2

189

Paper III Benthic O2 exchange in maerl and sand

Fig. 3

190

Paper III Benthic O2 exchange in maerl and sand

Fig. 4

191

Paper III Benthic O2 exchange in maerl and sand

Fig. 5

192

Paper III Benthic O2 exchange in maerl and sand

Fig. 6

193

Paper III Benthic O2 exchange in maerl and sand

Fig. 7

194

195

196

197