The Pennsylvania State University

The Graduate School

Department of Biology

THE ECOLOGY OF SEEP COMMUNITIES IN THE GULF OF MEXICO:

BIODIVERSITY AND ROLE OF LAMELLIBRACHIA LUYMESI

A Thesis in

Biology

by

Erik E. Cordes

Copyright 2004 Erik E. Cordes

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2004

The thesis of Erik E. Cordes was reviewed and approved* by the following:

Chuck Fisher Professor of Biology Thesis Advisor Chair of Committee

Katriona Shea Assistant Professor of Biology

Peter Hudson Willaman Professor of Biology

Michael Arthur Professor of Geosciences

Doug Cavener Professor of Biology Head of the Department of Biology

*Signatures are on file in the Graduate School

iii ABSTRACT

Cold seeps are common habitats along the continental margin in all the world’s oceans. In the Gulf of Mexico, they occur in the salt dome province of the upper

Louisiana slope, and along the base of the continental rise from Florida to Texas. Some of the most common inhabitants of cold seeps are vestimentiferan tubeworms which are entirely reliant on internal sulfide-oxidizing chemoautotrophic symbionts for their nutrition. The most common vestimentiferan tubeworm of the upper Louisiana slope is

Lamellibrachia luymesi. This, and other species of tubeworms, form aggregations of hundreds to thousands of individuals which harbor a diverse community. In this study, a total of 40 tubeworm aggregation and mussel bed samples containing at least 171 macrofaunal species were collected at seeps from 520 to 3300 m depth. The upper

Louisiana slope communities progress through a predictable sequence of successional stages. The youngest aggregations contain high biomass communities dominated by endemic species, with biomass decreasing over time as the relative abundance of non- endemic fauna in upper trophic levels increases. This process is mainly driven by the abundance of hydrogen sulfide in the epibenthic layer. Models support the hypothesis that L. luymesi alters its environment by releasing the sulfate generated by its internal symbionts into deeper sediment layers. This alters the distribution of sulfide leading to declines in sulfide concentration among the tubeworm tubes. The combination of these lines of evidence support the assertion that L. luymesi is a significant ecosystem engineer at hydrocarbon seeps in the Gulf of Mexico.

iv TABLE OF CONTENTS

LIST OF FIGURES ...... vi

LIST OF TABLES...... viii

ACKNOWLEDGEMENTS...... ix

Chapter 1 Introduction and A Review of Gulf of Mexico Cold Seep Ecology ...... 1

Introduction...... 1 Gulf of Mexico Cold Seep Ecology...... 5 Geological Setting ...... 5 Symbiotic Fauna ...... 14 Tubeworms...... 15 Mussels...... 18 Clams...... 20 Iceworms ...... 22 Community Ecology...... 23 Upper Slope Communities ...... 24 Deep Seep Communities ...... 32 Conclusions...... 34

Chapter 2 High Hydrogen Sulfide Demand of Tubeworm Aggregations Modifies the Chemical Environment at Deep-Sea Hydrocarbon Seeps ...... 36

Abstract...... 36 Introduction ...... 37 Methods ...... 39 Results ...... 43 Discussion...... 50

Chapter 3 Modeling the Mutualistic Interactions Between Tubeworms and Microbial Consortia...... 58

Abstract...... 58 Introduction ...... 58 Methods ...... 61 Population growth model ...... 62 Geochemical setting ...... 67 Model implementation ...... 73 Results and Discussion...... 79

Chapter 4 Succession of Hydrocarbon Seep Communities Associated with the Long-Lived Foundation Species Lamellibrachia luymesi...... 88 v Abstract...... 88 Introduction ...... 89 Methods ...... 92 Results ...... 98 Discussion...... 107

Chapter 5 Impact of an Ecosystem Engineering Tubeworm on the Habitat Characteristics and Community Structure of Gulf of Mexico Hydrocarbon Seeps...... 119

Abstract...... 119 Introduction ...... 120 Methods ...... 123 Results ...... 128 Discussion...... 134

Chapter 6 Community Structure of Gulf of Mexico Cold Seeps: Do General Theories of Deep-sea Ecology Pertain to Chemosynthetic Ecosystems?...... 141

Abstract...... 141 Introduction ...... 142 Materials and Methods ...... 146 Results ...... 152 Discussion...... 159

Bibliography ...... 169

Appendix...... 192

vi LIST OF FIGURES

Figure 2-1: Population growth characteristics of Lamellibrachia luymesi: Mortality rate ...... 44

Figure 2-2: Population growth characteristics of Lamellibrachia luymesi: Recruitment rate of aggregation BH7...... 45

Figure 2-3: Model output for Lamellibrachia luymesi aggregation BH7...... 46

Figure 2-4: Size-frequency of Lamellibrachia luymesi in aggregation BH7...... 49

Figure 2-5: Hydrogen sulfide uptake rate for whole Lamellibrachia luymesi aggregations...... 50

Figure 3-1: Model construction ...... 62

Figure 3-2: Growth model for L. luymesi...... 65

Figure 3-3:Concentration profiles of sulfate, sulfide, and dissolved organic carbon (DOC) ...... 69

Figure 3-4: Sediment porosity values ...... 74

Figure 3-5: Ratio of sulfide supply to sulfide uptake rate of L. luymesi aggregations based on known sources without sulfate release by tubeworm roots ...... 80

Figure 3-6: Ratio of sulfide supply to sulfide uptake rate of L. luymesi aggregations including sulfate release from tubeworm roots...... 82

Figure 3-7: Sources of sulfide available to tubeworm aggregations...... 83

Figure 4-1: Map of collection sites in the northern Gulf of Mexico...... 92

Figure 4-2: Multi-dimensional scaling (MDS) plot of community similarity ...... 101

Figure 4-3: Sulfide concentration in tubeworm aggregations...... 102

Figure 4-4: Changes in trophic structure with aggregation age...... 104

Figure 4-5: Similarity in species distribution among aggregations ...... 106

Figure 5-1: Map of collection locations in the northern Gulf of Mexico...... 124

Figure 5-2: Tubeworm growth models and estimated ages...... 129 vii Figure 5-3: Sulfide concentration within tubeworm aggregations...... 130

Figure 5-4: Correlations between predicted and measured hydrogen sulfide concentration...... 131

Figure 5-5: Multidimensional scaling plot of community similarity...... 133

Figure 6-1: Map of collection sites of communities associated with tubeworm aggregations or mussel beds ...... 147

Figure 6-2: Multidimensional scaling (MDS) plot of similarity among associated community samples from the upper Louisiana slope ...... 154

Figure 6-3: Multidimensional scaling (MDS) plot of similarity among associated community samples from all regions of the Gulf of Mexico...... 156

Figure 6-4: Diversity of communities associated with tubeworm aggregations and mussel beds throughout the Gulf of Mexico...... 157

Figure 6-5: Size of communities associated with tubeworm aggregations and mussel beds...... 158

Figure 6-6: Change in body size with depth of 5 common families of seep inhabitants...... 159 viii LIST OF TABLES

Table 2-1: Results of non-linear regression of number of Lamellibrachia luymesi recruits to population size...... 47

Table 2-2: Aggregation-specific recruitment period and comparison of model output to sampled size-frequency of Lamellibrachia luymesi...... 48

Table 3-1: Reported seepage rates for hydrocarbon and methane seeps ...... 71

Table 3-2: Parameters involved in diagenetic model...... 76

Table 4-1: Characteristics of collected tubeworm aggregations arranged by rank age...... 99

ix ACKNOWLEDGEMENTS

I would first like to acknowledge the informative discussions, support, and

camaraderie of the students of the Fisher Lab. I would not have begun this research without the inspiration from Derk Bergquist and John Freytag. I benefited from the

assistance of number people with the collections at sea including Stephane Hourdez,

Breea Govenar, Peter Deines, Sharmishtha Dattagupta, and Sue Carney. I have to thank a number of undergraduate students including Ben Predmore, Meredith Redding, Chris

Jones, Brian Tiegs, Michael McGinley, Liz Podowski, and Nicole Iacchei for assisting me with this research, and making sure that I didn’t take it too seriously. My research

would not have been possible without the assistance of the captain and crew of the RV

Seward Johnson II and the crew of the DSV Johnson Sea-link, and the captain and crew of the RV Atlantis and DSV Alvin. Bob Carney has been a sounding board for ideas in research and in my career and has enlivened our voyages to sea with wonderful stories. I have to thank Chuck Fisher for all of the guidance and inspiration along the way. I would like to acknowledge my parents, Ron and Cathy Cordes, who have always supported me no matter what I decided to do in my life. Finally, I have to thank Kristi Montooth for inspiring me in academia and in life.

Chapter 2 was previously published as Cordes et al. (2003) Hydrogen sulphide demand of long-lived vestimentiferan tube worm aggregations modifies the chemical environment at deep-sea hydrocarbon seeps. Ecol. Lett. 6: 212-219 and is reproduced here with permission from the publisher (Zoë Ellams, Permissions Coordinator,

Blackwell Publishing).

Chapter 1

Introduction and a Review of Gulf of Mexico Cold Seep Ecology

Introduction

Throughout the deep waters of the Gulf of Mexico lie numerous cold seeps; ecosystems based primarily on local productivity in the form of chemosynthesis.

Occupying these habitats are organisms with internal autotrophic symbionts that create complex physical structures in an otherwise relatively monotonous habitat. These biogenic structures support diverse assemblages of macrofauna including numerous endemic species along with colonists from the surrounding seafloor. Over the past 20 years, numerous investigations of the geology and geochemistry of the cold seeps along with investigations of the physiology and diversity of the seep inhabitants have led to a comprehensive description of these habitats. In the 6 chapters included here in my thesis,

I attempt to integrate the previously disparate data sets from geochemical and ecological investigations into a framework for understanding the underlying mechanisms for the patterns observed at the Gulf of Mexico cold seeps. While contributing to our knowledge

of the diversity of fauna inhabiting the seeps of the Gulf of Mexico, I also examine the

role of Lamellibrachia luymesi in structuring these communities through its influence on

the biogeochemistry of the upper Louisiana slope seeps.

The first chapter is a review of the previous knowledge of Gulf of Mexico seep

ecology. This is intended to provide the reader with the necessary background to

2 understand the significance of the problems addressed in later chapters. It summarizes the geology of the Gulf of Mexico including the various sources of reduced chemicals that provide the energy necessary to sustain the seep communities. The common symbiotic fauna of the seeps are presented with respect to their physiology and ecology. Finally, our existing knowledge of the community ecology of the seeps is summarized, including the most significant findings of the subsequent chapters.

In Chapter 2, a population growth and sulfide uptake model for Lamellibrachia luymesi is presented This model has allowed the estimation of L. luymesi aggregation age that is used in subsequent chapters, and in other ongoing studies. The model also provides estimates of the sulfide uptake rate from seep sediments by intact tubeworm aggregations, a question that would be intractable by normal experimental means. This study estimates sulfide uptake rates at over 30 mmol·hr-1. It is further speculated that this

high rate of sulfide removal from seep sediments may influence the distribution of sulfide

in the tubeworm habitat, and may influence the structure of the macrofaunal community

inhabiting the aggregations.

In Chapter 3, an extension of the model presented in Chapter 2 is used to test

some of the predictions made about the sulfur cycle in tubeworm-influenced sediments.

Specifically, it tests that hypothesis that L. luymesi releases the sulfate generated by its internal sulfide-oxidizing symbionts into seep sediments to enhance sulfide generation.

Model results suggest that the release of sulfate into organic-rich sediments is necessary to generate sufficient sulfide to support large aggregations of L. luymesi. This finding alters our existing concept of the biogeochemistry of tubeworm-influenced sediments and

3 demonstrates the profound influence of L. luymesi on the characteristics of the habitat it creates.

The potential influence of L. luymesi on the macrofaunal community inhabiting the tubeworm aggregations is examined in Chapters 4 and 5. In Chapter 4, a previously proposed succession model is tested in L. luymesi aggregations from GC234 and GC232.

Quantitative age estimates are generated for each aggregation to place them in a temporal sequence. The succession model is validated, with biomass and sulfide concentration declining over time along with shifts in trophic structure from communities dominated by endemic species in lower trophic levels to those dominated by non-endemic species in higher trophic levels.

In Chapter 5, we examine the successional changes in community structure with respect to the effects of L .luymesi on the sulfide distribution within the habitat structure it creates. Though sulfide concentration generally declines over time, this rate of decline is different in aggregations of varying proportions of L. luymesi. Sulfide is generally lower in older aggregations with more L. luymesi. Though causation is difficult to prove, the previous modeling studies presented in Chapters 2 and 3 provide a mechanistic explanation for these findings. Sulfate release into deeper sediment layers by L. luymesi could have the effect of shifting the site of sulfate reduction away from the sediment- water interface while enhancing carbonate deposition. Together with their high rates of uptake from surficial sediments, this would have the effect of lowering sulfide in L. luymesi dominated aggregations, as was observed in this study. This demonstrates the ecosystem-engineering capacity of this tubeworm and the impact it has on the surrounding benthos.

4 In Chapter 6, community data from 40 collections obtained over the past 7 years is integrated to form a cohesive picture of the biodiversity and ecology of the cold seeps of the Gulf of Mexico. A total of 162 species were identified, greatly increasing the known diversity of cold seep inhabitants world-wide. Changes in the structure of these communities with respect to the depth of collection are examined. While diversity decreased and endemicity increased with depth, abundance and biomass of associated fauna showed little response to depth. This is likely due to the reliance of the endemic fauna on local productivity that does not decline with depth, in contrast to the quantity and quality of photosynthetically-derived production.

These studies have uncovered previously unknown diversity at the cold seeps of the Gulf of Mexico, and have verified existing hypotheses concerning succession processes within these communities and the release of sulfate through the roots of L. luymesi. However, it is apparent that during the course of these investigations, more questions were posed than were answered. It is my hope that these studies will influence the direction of research in the Gulf of Mexico by providing numerous testable hypotheses for future investigations.

5 Gulf of Mexico Cold Seep Ecology

Cold seeps are common deep-water habitats on the continental slope and rise world-wide. They occur from the coast of California and Oregon to the Aleutian subduction zone and Japan across the northern Pacific and from the Lau Basin to the coasts of Chile and Costa Rica in the southern Pacific. In the Indian Ocean, seeps are known from the coasts of Pakistan and Mauritiana. In the Atlantic they occur near

Barbados, the Blake Ridge off the Carolinas, the coast of Norway, the Mediterranean, and the coast of Algeria. The hydrocarbon seeps of the Gulf of Mexico (GoM) contain some of the most well described seep communities in the world. The cold seeps of the

Gulf of Mexico were first discovered in the mid-1980s along the Florida Escarpment

(Paull et al. 1984). Additional shallower seep communities were discovered by a group of geologists surveying what they thought would be the depauperate communities overlying potential hydrocarbon reserves on the upper Louisiana slope (Kennicutt et al. 1985).

Since then, GoM seeps have been the subject of numerous research programs describing the geology and geochemistry of the habitat and the physiology and ecology of the autotrophic and heterotrophic fauna. Here, we endeavor to summarize the current state of knowledge of the seeps of the Gulf of Mexico, focusing on the ecology of the symbiotic organisms inhabiting the seeps and the diverse heterotrophic fauna associated with them.

Geological Setting

The cold seeps of the Gulf of Mexico are largely driven by the process of salt

tectonics caused by an underlying salt layer. This layer was formed by the successive

6 opening and closing of the GoM over the last 200 million years (Pindell 1994). The

GoM was originally formed during the late Triassic (Sassen et al. 2001), as the North and South American plates separated. The newly formed basin was filled with seawater from what would eventually become the Atlantic Ocean. In the middle

Jurassic, the opening to the Atlantic was closed (Pindell 1994). Extensive evaporation followed leaving behind a thick salt sheet, the Louann salt formation, on the basement of the GoM. Once the basin was reopened, it refilled with seawater and evaporite deposition ceased. Additional evaporite formation appears to have occurred approximately 55-50 million years ago during the Paleocene and Eocene as the

Caribbean plate moved to the east across the region (Rosenfeld and Pindell 2003).

When the Cuban Arc impacted the Yucatan and North American plates, it formed a temporary barrier accompanied by a rapid drop in sea level. By the early Miocene, approximately 21 million years ago, the Yucatan Strait opened allowing water to flow into the GoM from the newly formed Caribbean Sea.

Around the same time, deposition of large volumes of siliciclastic sediments onto the salt layer began (Sassen et al. 1994). As these sediments accumulated they compressed and formed a thick layer of sandstone and shale over the salt basement

(McGookey 1975). These impermeable layers form a barrier preventing the further migration of hydrocarbons from deeply buried Mesozoic source rocks (Kennicutt et al. 1992). Differential deposition of sediments on the continental shelf caused the salt sheet to migrate down along the continental slope (Humphris 1979). This deformed the incompressible salt sheet into vertically mobile pillars and salt domes (Humphris

7 1979). Salt diapirs subsequently pierced and cracked the shale and sediment

overburden, causing the migration of trapped hydrocarbons (Kennicutt et al. 1988).

Much of the oil becomes trapped in extensive reservoir quality sands of late Pliocene

to early Pleistocene origin (Milkov and Sassen 2003), though some of it reaches the

seafloor at the hydrocarbon seeps. This process of salt tectonics dominates the

northern Gulf of Mexico, particularly in the areas of most intensive seepage on the

upper Louisiana slope (ULS).

What follows is a description of our current state of knowledge of the types of seeps in the Gulf of Mexico with examples from the known sites. This information is derived from submersible observations, trawl data, and sediment samples. Compiling this information into a picture of the benthic habitats of the GoM is much like attempting to describe a forest from a helicopter using only a flashlight. Various descriptions are integrated to the best of our ability, though we are constantly adding to the pool of information on the GoM, and much of what is included below was gleaned from discoveries made on exploratory expeditions over just the last 3 years.

There are three main types of surface manifestation of seepage along the ULS: mud volcanoes, brine seeps, and oil seeps (Sassen et al. 1994, Aharon and Fu 2000). The most vigorous of these are the mud volcanoes with active venting of subsurface fluids

(MacDonald et al 2000). The fluids migrate along vertical faults in the overlying sediments and maintain distinct temperature anomalies to the seafloor (MacDonald et al

2000). Venting fluids cause a high degree of sediment instability, limiting or precluding the attachment of most of the seep fauna that require hard substrata for settlement

8 (Roberts and Neurauter 1990). Rather, the fauna of the mud volcanoes is dominated by bacterial mats and mobile chemosynthetic clams, though mussel beds have been observed

(MacDonald et al. 2000). These sites are known from Green Canyon (GC) 143 (Roberts and Neurauter 1990), GC286 (Sassen et al. 2003), Garden Banks (GB) 424/425

(MacDonald et al. 2000), and Mississippi Canyon (MC) 697 (Prior et al. 1989)1.

Brine seeps of the upper Louisiana slope are sites of moderate subsurface fluid migration. They do not have the characteristic rapid fluid venting of the mud volcanoes,

rather the fluids “seep” through the sediment layers. Surface manifestation of brine seeps

may be limited to small, darkly colored depressions, or the dense seeping fluids may

accumulate in large pools and rivers along the sediment surface (MacDonald et al.

1990a). The brine is highly saline, with salinities up to 130 ppt, and is supersaturated with

methane (MacDonald et al. 2000). The abundance of methane in the brines leads to the

presence of bacterial mats and occasionally mussels with methanotrophic symbionts,

Bathymodiolus childressi.

One of the most striking brine seeps of the ULS is Brine Pool NR1. This was a

pockmark likely formed from the rapid sublimation of a large volume of gas hydrate

(MacDonald et al. 1990a). The brine originates from a shallow (<500 m) subsurface salt

diapir and collects in the depression formed by the blowout event. The pool is

1 The names of the various sites are derived from the partitioning of the U.S. waters of the Gulf of Mexico into oil rights lease blocks by the Minerals Management Service. Lease blocks are designated with numbers from east to west and north to south within lease areas. The lease areas discussed in this study are Garden Banks (GB), Green Canyon (GC), Mississippi Canyon (MC) Vioska Knoll (VK), Alaminos Canyon (AC), Atwater Valley (AT), and Florida Escarpment (FE).

9 approximately 22 m long and 11 m wide and is surrounded by a wide band of B. childressi. Most other brine seep sites are dominated by more diffuse surface expressions of seepage. GC204 contains a site of rapid fluid transport related to active salt tectonism

(Sassen et al. 2003). The high concentration of methane in the brine at this site leads to the formation of methane hydrate. Abundant methane provides the energetic resources for extensive mussel beds at this site, though not at the same scale as the brine pool. In

Mississippi Canyon (MC) 929, active seepage of brines containing elevated metal concentrations leads to the formation of barite chimneys (Chapter 6, pers. obs.). Though relatively toxic and moderately radioactive, these chimneys harbor high biomass B. childressi communities as well. More diffuse brine seeps are found in the Garden Banks area (GB535, 543, 544) which is characterized by relatively thin authigenic carbonate crusts on the sediment surface (Sager et al. 2003, Chapter 5, 6). The known mussel populations at the GB sites are confined to a few individuals inhabiting the structure provided by small tubeworm aggregations.

Oil seeps are characterized by the slowest seepage rate of subsurface fluids. The oil accumulates beneath low porosity traps in the stratigraphy and seeps along faults created by the impact of mobile salt pillars and diapirs (Kennicutt et al. 1992). While they may occasionally be associated with active bubble streams of methane and oil droplets emerging from the sediment surface, they are more commonly limited to oil rich sediments with little apparent surficial differentiation from the surrounding sediments

(Sassen et al. 1994). The abundance of tubeworms at these sites is likely due to slower seepage rates of hydrocarbons leading to an increased amount of time for subsurface hydrocarbon degradation by microbes coupled to the production of hydrogen sulfide

10 (Sassen et al. 1994, Chapter 3). However, the chemosynthetic and heterotrophic fauna of oil seeps is variable depending on the relative abundance of gaseous and liquid hydrocarbons which changes over time (Chapter 4).

One of the most concentrated areas of oil seepage on the ULS is in an area of active hydrocarbon extraction in the Green Canyon (GC) oil lease block. Two of the best known sites in this area are GC185, also known as Bush Hill, and GC234. GC234 lies on top of a fault scarp that runs roughly east-to-west (MacDonald et al. 2003). The broad area impacted by this structure has provided ample suitable substrata for the development of chemosynthetic communities as well as gorgonian and scleractinian coral colonies.

There are a number of distinct mussel beds, and contiguous tubeworm aggregations covering hundreds of m2. This fault scarp extends into adjacent lease blocks (MacDonald

et al. 2003), and re-emerges at the sediment surface at GC232, forming an additional

chemosynthetic community of somewhat lesser magnitude. Bush Hill is just north of

these two sites, and lies on top of a large salt dome (MacDonald et al. 1989). At the crest

of the hill are numerous distinct tubeworm aggregations and mussel beds. Around the

periphery are hard substrata with gorgonian (Callogorgia americanus) and scleractinian

(Lophelia pertusa) colonies.

The site at GC354 appears to be similar in geological structure to the Bush Hill

site (Chapter 5, 6). This site lies at the top of a knoll approximately 30 km to the west of

the central ULS sites. The “hill” at this site is of similar magnitude, though the crest rises

to 520 m, 150 m above the base of the mound. At the crest are colonies of Callogorgia

americanus and Lophelia pertusa, with tubeworm aggregations occupying the flanks and base of the mound, somewhat in reverse of the habitat characteristics of Bush Hill. The

11 oil seeps in Mississippi Canyon (MC885) and Vioska Knoll (VK826), contain large, roughly rectangular carbonate blocks that appear to be the result of recent uplift events rather than ongoing authigenic carbonate precipitation (Schroeder 2002, Chapter 6).

Tubeworm aggregations are associated with many of these blocks where sulfide-rich fluids migrate along the subsurface channels associated with their uplift. Along with areas of Bush Hill, GC234 and GC354, these two sites share the common feature of containing coral colonies along with tubeworm aggregations, often occupying the same carbonate outcrop.

Gas hydrate accumulations are often associated with both brine and oil seeps.

Hydrates are ice-like clathrates of methane gas within a water molecule lattice (Sassen et al. 2001). They form under high pressure and low temperature conditions at depths between 440 and 2400 m in the northern Gulf of Mexico (Sassen et al. 1999). These structures may crystallize in the pore spaces of sediments forming small nodules, or may accumulate as massive vein filling structures where gas migration is most rapid (Milkov and Sassen 2003). Mussel beds, bacterial mats, and the iceworm Hesiocaeca methanicola

(Fisher et al. 2000) are often associated with outcroppings of methane hydrate.

Tubeworm aggregations can also occupy the sediments overlying gas hydrates. The frequent occurrence of chemosynthetic communities within the zone of greatest hydrate abundance has led some authors to suggest that the concentrated methane in hydrates enhances sulfate reduction by microbial consortia producing a persistent and sufficient sulfide source to fuel these communities (Carney 1994).

Hydrocarbon seepage on the lower slope and continental margin results from

axillaries of the general phenomenon of salt tectonics. The Florida Escarpment is a

12 massive carbonate wall at the base of the continental plate off the western edge of

Florida. It rises from 3300 m to a depth of approximately 2000 m. Where the base of

this wall meets the abyssal plain, there are extensive areas of hypersaline brine

seepage hypothesized to result from dewatering of the Florida platform (Paull et al.,

1984). The presence of biogenic sulfide (Cary et al. 1989) in addition to geothermal

sulfide suggests that the seepage along the escarpment may be a result of deep seated

sediment compression and mobility of the brine along the intersection of the platform

and abyssal plain sediments.

The Sigsbee escarpment at the base of the Louisiana slope is similar in form, but of very different composition. In this area, the large cliffs emerging from the abyssal plain at approximately 3000 m are composed of salt rather than carbonate (Bryant et al.,

1990). Seepage along this escarpment is produced by the continuing encroachment of the

Luann salt sheet onto the abyssal plain. The presence of the salt in shallow subsurface layers forms numerous brine pools and flows along the lower slope (Brooks et al. 1990).

One of the largest pools of brine is the Orca Basin on the lower slope south of the

Mississippi Canyon (Brooks et al. 1990). This basin is filled with hypersaline brine which goes through successive density layers as it mixes with the overlying seawater. This leads to a gradual halocline with increasing depth from 2200 to 2400 m inside the basin. The broad hypoxic and hypersaline zone within the basin precludes the colonization of most organisms, the exception being numerous species of sponges and an occasional holothurian, Benthodytes typica (Brooks et al. 1990).

13 Above the Sigsbee Escarpment is Atwater Valley, an extension of the

Mississippi Fan. Atwater Valley lies at the base of the Mississippi Fan within the salt

dome province and exhibits similar seepage patterns to the Louisiana slope (Milkov

and Sassen 2003). The site at AT425 is a large mound that lies atop a salt ridge with

active areas of seepage along the crest and flanks (MacDonald et al. 2003). The flanks

are dominated by a series of high relief ridges running from the crest to the base of

the knoll. The peaks of each of these ridges contain areas of seepage supporting

mussel beds and the occasional vestimentiferan (Escarpia laminata) aggregation

(Chapter 6).

Alaminos Canyon is formed by a cleft in the Louann salt sheet on the continental shelf of Texas. This canyon cuts through the continental shelf between 2200 and 3000m exposing alternating salt and carbonate layers (Bryant et al., 1990). The most extensive chemosynthetic communities (B. brooksi and B. childressi mussel beds and Escarpia laminata tubeworm aggregations) in Alaminos Canyon are associated with terraces along the sides of the canyon formed by lateral salt “tongues” (Brooks et al. 1990, Liro 1992).

On the tops of the terraces are uplifted carbonate blocks in addition to presently forming authigenic carbonates. Seepage is in the form of active methane bubble streams that can occasionally form methane hydrates (Chapter 6).

A recent exploration of the continental margin west of the Yucatan Peninsula

discovered an additional type of seepage in the Gulf of Mexico (MacDonald 2004).

The Campeche Knolls are a series of salt diapirs that rise up from a depth of 3000 m.

Along the crest and flanks of these knolls are extensive areas of asphalt volcanism.

14 The asphalt appears to be discharged under high temperature and then undergoes

rapid cooling upon contact with the seawater. In the sediments associated with these

areas were liquid petroleum, gas hydrate, and gaseous hydrocarbons supporting a

diverse chemosynthetic and heterotrophic fauna.

Symbiotic Fauna

Associated with the hydrocarbon and methane seeps of the GoM are three groups

of fauna symbiotic with chemosynthetic of methanotrophic bacteria; tubeworms,

mussels, and clams. These metazoans contain internal microbial symbionts that utilize

sulfide (tubeworms, mussels, and clams), methane (mussels), or both (mussels) as an

energy source. There are currently 5 or 6 species of vestimentiferan tubeworms, 5 or

6 species of modiolid mussels, and at least 5 species of lucinid, thyasirid, and

vesicomyid clams that are known from the various seeps in the Gulf of Mexico. The

recent exploration of the Campeche Knolls (3000 m depth) in Mexican territorial

waters of the Gulf of Mexico (MacDonald et al. 2004) discovered additional

chemosynthetic species (1 lamellibrachid tubeworm, 1 modiolid mussel, 1

vesicomyid clam, 1 solemyid clam) that are not included in the discussion here due to

their unresolved taxonomic status and limited knowledge of their ecology.

15 Tubeworms

Vestimentiferan tubeworms were originally described as a separate phylum affiliated with the Phylum Pogonophora. Recent work has placed these two groups together within the Family Siboglinidae in the polychaete Order Sabellida (Rouse 2001).

Within the siboglinids there are three clades of tubeworms, the lamellibrachids and escarpids inhabiting seeps, and the alaysids encompassing all of the hydrothermal vent species. In the Gulf of Mexico, there are 2 or 3 species of lamellibrachids:

Lamellibrachia luymesi (Gardiner and Hourdez 2003), Lamellibrachia sp. nov. 1 (Nelson and Fisher 2000), and a lamellibrachid recently collected at 1880 m in Atwater Valley

(Chapter 6). It is still unknown at this time whether L. sp. nov. 1 and the new lamellibrachid are the same species. There are 3 species of escarpids in the GoM:

Escarpia laminata (Jones 1985) from the seeps at the Florida Escarpment, Alaminos

Canyon and Atwater Valley, Seepiophila jonesi (Gardiner et al. 2003) from the upper

Louisiana slope, and an undescribed species of escarpid from the upper Louisiana slope

(McMullin et al. 2003).

All vestimentiferans lack a digestive tract as adults and are entirely reliant on internal, sulfide-oxidizing bacteria for their nutrition (Fisher 1990). The mode of transfer of nutrients to the host has been proposed to involve both intracellular digestion of symbionts (Bosche and Grassé 1984) and leaching of secondary metabolites by the symbionts (Felbeck and Jarchow 1998, Bright et al. 2000). Symbionts are acquired in the larval stage of each successive generation with the symbiont strain determined more by location than host species (Nelson and Fisher 2000, McMullin et al. 2004).

16 The bacteriocytes containing the symbionts are housed in an internal organ, the trophosome. Trophosome tissue is highly vascularized with the vascular system containing blood capable of reversibly binding both oxygen and sulfide (Jones 1985, Arp and Childress 1981). Hydrothermal vent tubeworms, such as Riftia pacyptila, obtain both oxygen and sulfide across the plume, acquiring the two competing chemicals at different times as they encounter the plume surface (Arp et al. 1987). Unlike the majority of vestimentiferans, L. luymesi acquires sulfide through a posterior extension of its body and tube, the root (Julian et al. 1999, Freytag et al. 2001). Laboratory studies have shown that these tubeworms can obtain sufficient sulfide across their roots to fuel net autotrophy, taking up carbon dioxide across their plume (Freytag et al. 2001). This allows L. luymesi to separate the acquisition of oxygen and sulfide spatially.

The properties and relative amounts of the hemoglobins present in the coelomic and vascular blood of the ULS species are highly specialized and species-specific

(Freytag 2003). The blood of L. luymesi contains hemoglobins with relatively low affinity but high capacity for sulfide, allowing it to bind large amounts of the consistently high concentrations of sulfide in the sediments around its roots. S. jonesi has blood with higher affinity but lower capacity hemoglobins, specializing in more ephemeral sulfide sources in lower concentrations present in the epibenthic layer around its plume. This potentially allows the coexisting species of tubeworms to exploit different microhabitats within the seep environment (Freytag 2003).

As surface expression of seepage and the sulfide generated by the processes it fuels subsides, the importance of L .luymesi roots as a site for sulfide uptake increases

(Chapter 3). From the recruitment data and existing measurements of water column

17 sulfide levels, it appears that sulfide is only rarely present around the plumes of L. luymesi after 60 years (Bergquist et al. 2003b, Chapter 2, 5). This suggests that all of the sulfide requirements for older L. luymesi aggregations are met by subsurface sulfide pools. The majority of sulfide produced in seep sediments is believed to be a result of anaerobic methane oxidation by microbial consortia, with the rates of anaerobic methane oxidation at ULS seeps among the highest rates ever measured (Aharon and Fu 2000).

Sulfide is also produced by the anaerobic oxidation of deeply buried organic material and the interaction of methane rich brines with subsurface mineral deposits of gypsum and anhydrite (Carney 1994, Sassen et al. 1994, Saunders and Thomas 1996). These sources amount to the sum of sulfide present in seep sediments away from the influence of tubeworm aggregations.

In models of sulfide supply to L .luymesi aggregations, the combination of these sources is not sufficient to match the high sulfide demands calculated (Chapter 2, 3). A likely candidate for an additional source of sulfide is the sulfate generated by the tubeworm’s symbionts (Julian et al. 1999, Freytag et al. 2001). By releasing this sulfate through their roots into seep sediments, the tubeworms would supply an energetically favorable oxidant in deeper sediment layers. In the models, the increase in sulfide production resulting from the additional sulfate supplied was sufficient to match the demands of even large tubeworm aggregations (Chapter 3). This theoretical study generated numerous hypotheses that await empirical tests to validate the presence and significance of this complex interaction among symbiotic tubeworms and microbial consortia.

18 Mussels

There are at least 5 species of modioliform mussels with chemosynthetic and/or

methanotrophic symbionts in the Gulf of Mexico; Bathymodiolus childressi from the

upper Louisiana slope and Alaminos Canyon, B. brooksi from Alaminos Canyon and

the Florida Escarpment, and B. heckeri from the Florida Escarpment, Tamu fisheri

from the upper Louisiana slope, and Idas macdonaldi from the upper Louisiana slope

(Gustafson et al. 1998). The mussels recently collected in Atwater Valley

(MacDonald et al. 2003, Chapter 6) are morphologically and genetically similar to B.

brooksi (S. Carney, pers. comm.), though their taxonomic status remains unresolved.

B. childressi from ULS and Alaminos Canyon were originally considered two

separate species (Seep Mytilid Ia and Ib, Fisher 1993, Nelson and Fisher 1995) and

genetic differentiation between the two populations approaches that of separate

species (Craddock et al. 1995), though they currently regarded as conspecific

(Gustafson et al. 1998).

Mussels inhabiting vents and seeps exhibit the widest variety of symbiont

types and levels of symbiont integration of the three main groups of symbiotic fauna.

Of the mussels of the Gulf of Mexico, Bathymodiolus childressi is the best studied

and therefore most of the information below is derived from information on their

biology. They contain type I methanotrophs in their gill tissues that obtain energy

through the oxidation of methane and incorporate methane carbon into organic

compounds (Childress et al. 1986, Fisher et al. 1987). B. childressi are capable of

growing with methane as their sole carbon and energy source (Cary et al. 1988),

19 leading to extremely light carbon stable isotopic signatures in the -40 to -70 range

(Childress et al. 1986, Brooks et al. 1987, MacAvoy et al. 2002). They are still capable of filter feeding (Page et al. 1990), though they appear to obtain most of their nutrition from the symbionts (Fisher and Childress 1992), and contain hypertrophied gills and reduced digestive tracts (Gustafson et al. 1998). Bacteriocytes are clustered towards the outer edges of the gill filament tissues (Fisher et al. 1987), which likely facilitates diffusion of methane and oxygen from the gill surface to the symbionts

(Nelson and Fisher 1995). Diffusion is the sole mode of transport of dissolved gasses as the mussels do not contain binding proteins for transport of methane or oxygen

(Childress and Fisher 1992). Transfer of nutrients from the symbionts to the host is largely accomplished through intracellular digestion of the symbiont rather than translocation of secondary metabolites (Streams and Fisher 1997). In Bathymodiolus species on the Mid-Atlantic Ridge, symbionts are acquired each successive generation, apparently through pits in the membranes of cells lining the gill surface

(Won et al. 2003).

Far less information has been published on the other species of mussels from the Gulf of Mexico. B. brooksi contains both methanotrophic and sulfide-oxidizing symbionts. These symbionts may occur within the same cell, often within the same vacuole (Fisher et al. 1993). The two different types of symbionts can occur in different relative abundance in populations inhabiting different microenvironments, with some B. brooksi found in areas of Alaminos Canyon with up to 222 µM sulfide

(Fisher et al. 1993). B. heckeri contains symbionts with two different morphologies in

20 bacteriocytes lining the gill filaments (Cavanaugh et al. 1987). If in fact different

strains, these symbionts may not represent different physiological capacities, since

only methanotrophy has been demonstrated in this species (Cary et al. 1989, Fisher

1993). T. fisheri distribution is restricted to the ULS, inhabiting the base of L. luymesi

aggregations and mussel beds (Fisher 1993, Chapter 4). T. fisheri contains sulfide-

oxidizing symbionts (Fisher 1993) contained in extracellular gill “pockets” (Craddock

et al. 1995). It also harbors a commensal polynoid polychaete, Branchipolynoe

symmitilida, which is also commonly found in B. heckeri and the hydrothermal vent

species B. thermophilus (Fisher 1993). Very little is known of the other GoM seep

mytilid, Idas macdonaldi, as it has only been collected from a seep site in GB386

(Gustafson et al 1998). Other members of the genus Idas are associated with the

sulfide rich deep water habitats of wood falls and whale carcasses (Distel et al. 2000).

Clams

There are three groups of clams inhabiting the seeps of the Gulf of Mexico.

On the upper Louisiana slope, there are at least two species of vesicomyids

(Vesicomya chordata and Calyptogena ponderosa), two species of lucinids

(Lucinoma atlantis and an undescribed species of Lucinoma) and one thyasirid

(Thyasira oleophila). Another lucinid and another thyasirid have been collected at

both the Alaminos Canyon and Florida Escarpment sites (Chapter 6), but their

taxonomic status remains unresolved. In addition, unidentified vesicomyids have

been reported from the Florida Escarpment (Paull et al. 1984).

21 All of these bivalves are chemoautotrophic, exhibiting a variety of morphological adaptations to their symbiotic life style. In the vesicomyids

Calyptogena ponderosa and Vesicoyma cordata, the gut and labial palps have been greatly reduced (Fisher 1990). All described vesicomyids harbor intracellular sulfide- oxidizing symbionts in bacteriocytes in greatly enlarged gills (Brooks et al. 1987).

They contain abundant extracellular hemoglobin in a closed circulatory system (Scott and Fisher 1995) which is capable of concentrating sulfide 2 to 3 orders of magnitude above environmental levels (Fisher 1996). Sulfide is taken up across the surface of the foot which is buried in the sediments, and oxygen taken up through the siphons extended into the water column. These clams burrow through the sediment, leaving characteristic trails behind. They appear to pursue ephemeral sulfide sources (Fisher

1990), chasing often undetectable seepage across the sedimented habitat (Scott and

Fisher 1995). The ULS clams are rarely collected alive, more commonly observed are the lebenspurren of clams leading to beds of shell hash where the sulfide trials ran out.

The lucinids and thyasirids have a reduced gut and feeding palps and lack an incurrent siphon. They appear to obtain sulfide through extensions of their foot, which forms a “feeding tube” (Turner 1985). These clams form extensive sub-surface burrows in order to acquire sulfide from deeper anoxic sediment layers (Turner

1985). It has been suggested that some lucinids oxygenate their burrows and mobilize the sulfide from pyrite deposits, thereby “mining” sulfide from the sediments (Dando et al. 1994). Some thyasirids are capable of extending their foot over 30 times the

22 length of the shell, potentially mining sulfides from anoxic layers of sediment

(Dufour and Felbeck 2003). Lucinids contain intracellular bacterial symbionts housed

in bacteriocytes in the gills (Fisher 1990). Unlike the majority of symbiotic bivalves,

the symbionts of thyasirids are extracellular (Southward 1986, Fisher 1990). Nutrient

transfer from symbiont to host appears to be through intracellular digestion (Distel

and Felbeck 1987), though this has not yet been verified for all species.

Environmental symbiont transmission has been shown in Lucinoma aequizonata and

all other lucinids examined to date (Gros et al. 1999).

Iceworms

An additional group considered here are the iceworms (Methanoaricia

dendrobranchiata, Desbruyères and Toulmond 1998). Iceworms are hesionid

polychaetes that inhabit the surface of methane hydrates on the upper Louisiana slope

(Desbruyères and Toulmond 1998). They were first discovered living in densities up

to 2500 individuals per m2 within small depressions carved into an outcropping of

methane hydrate (Fisher et al. 2000). The iceworms apparently create these

depressions by fanning their parapodia and increasing the water movement over the

surface of the clathrate. This may facilitate the sublimation of the hydrate and

increase the dissolved methane and sulfide available to free living microbes from

which they likely derive the majority of their nutrition (Fisher et al. 2000).

23 Community Ecology

Along with the symbiotic species of the Gulf of Mexico are a total of 171

species (retained on a 2 mm sieve) associated with tubeworm aggregations and

mussel beds (Chapter 6). Many of these species are representatives of groups of

animals that are found at cold seeps and hydrothermal vents worldwide. It is

commonly believed, and occasionally demonstrated, that these endemic species have

developed a suite of adaptations that allow them to persist in an otherwise

exclusionary environment of high environmental toxicity (Fisher et al. 2000, Hourdez

et al. 2002). However, these are not the only organisms inhabiting the biogenic

habitat created by the symbiotic organisms at the seeps of the Gulf of Mexico. The

rest of the fauna have been dubbed “colonists” when occurring in greater numbers at

the seeps or “vagrants” when occurring in similar low densities comparable to non-

seep habitats (Carney 1994). These seep inhabitants may be taking advantage of the

local primary production in these deep-sea chemosynthetic habitats, or merely

occupying the complex habitat provided by the foundation species of tubeworms and

bivalves (Fisher 1993, MacAvoy et al. 2002). The majority of the information on the

ecology of Gulf of Mexico seeps comes from the ULS. A working model of

succession has been proposed (Bergquist et al. 2003a) and validated (Chapter 4) at

these seeps. The deeper seeps of the GoM are less well known, with their ecology

remaining in the descriptive stage. These two habitats will be considered separately.

24 Upper Slope Communities

Quantitative samples of the tubeworm and mussel associated communities of the

ULS have been obtained on a total of 8 cruises over the past 9 years. The majority of this sampling effort has been focused on a cluster of 4 sites in the Green Canyon oil lease block: Bush Hill, Brine Pool, GC234 and GC232. The abundance of geochemical data from these sites has allowed a comprehensive description of the temporal evolution of abiotic conditions at the seeps and coincident changes in the composition of the associated community. The conclusions drawn from these central sites have been expanded in recent years to include the biodiversity found at 8 additional sites across 400 km on the upper slope of the northern Gulf of Mexico. These sites generally occur under similar geochemical regimes, though the relative age of the site and the amounts of methane and sulfide present in the habitat is variable.

Early in the evolution of a seepage source, seeping fluids contain high concentrations of methane (Bergquist et al in press). This is evidenced by the presence of active methane bubble streams and oil droplets emanating from these sites. High methane concentrations result in extremely high rates of anaerobic methane oxidation, among the most rapid measured in any environment (Aharon and Fu 2000). The increase in alkalinity resulting from anaerobic methane oxidation results in the precipitation of authigenic carbonates (Boetuis et al. 2000, Aharon and Fu 2000). Carbonate provides the necessary hard substrate for settlement of vestimentiferans and mussels alike (Tyler and

Young 1999). Though these species likely settle concomitantly, the preponderance of methane at young seeps favors the establishment of methanotrophic mussel populations.

25 Mussels will not only dominate the space available, but are capable of filter feeding and reducing the numbers of larvae reaching the substrate. This may contribute to an initial phase of low levels of tubeworm recruitment into mussel-dominated nascent seepage sites.

The associated community of mussel beds is dominated by endemic species.

The proportion of species that were considered to be endemic ranges from 60 to

100% in individual samples (Bergquist et al in press). Some endemic species have

been shown to be tolerant of hypoxia and elevated methane and sulfide

concentrations. The previously mentioned iceworm, Hesiocaeca methanicola, is able

to persist in anoxia in excess of 96 hours (Fisher et al. 2000). An orbiniid polychaete,

Methanoaricia dendrobranchiata, commonly found in mussel beds, contains high

affinity hemoglobins and enlarged gills allowing it to survive for days in anoxic

conditions in the presence of 1 mM concentrations of sulfide (Hourdez et al. 2002).

The gastropod Bathynerita naticoides is a common mussel associate with fungal

symbionts that are suggested to assist in environmental detoxification (Zande 1999).

These adaptations allow these and other endemic species to exploit the locally

elevated primary production in the form of chemosynthetic microbial biomass in a

habitat with reduced risk of predation imposed by higher trophic levels (Bergquist et

al. 2003a, Chapter 4). The limited diversity of the mussel-associated community (a

total of 19 species in 17 collections, Bergquist et al in press) suggests that while there

are advantages to the colonization of this harsh environment, it requires a relatively

rare set of adaptations. The relatively low number of families of animals which have

been able to colonize mussel beds and young tubeworm aggregations, including the

26 hesionidae, polynoidae, bresilidae, neritoidae, provannidae, and galatheidae (Sibuet

and Olu 1998, VanDover 2000) also suggests that the evolution of this tolerance has

been a rare event However, the species-level diversity represented within these

groups suggests that once obtained this set of adaptations may allow these families to

rapidly diverge and colonize a range of habitats at cold seeps and hydrothermal vents.

Overlapping with and persisting past the mussel bed successional stage are young tubeworm aggregations. During this period, anaerobic methane oxidation and carbonate precipitation rates remain high (Aharon and Fu 2003, Chapter 3). Additional carbonate precipitation will allow additional space for vestimentiferan and mussel larvae to settle.

Increased tubeworm settlement rates persist for the first 10 to 30 years of aggregation development (Bergquist et al. 2002, Chapter 2). Model results indicate that active recruitment coupled with lower rates of mortality for those individuals which have attained a certain size results in rapid increases in tubeworm population size (Chapter 2).

L. luymesi, along with Seepiophila jonesi and the undescribed escarpid species, form aggregations of hundreds to thousands of individuals on the upper slope of the Gulf of

Mexico (Kennicutt et al. 1995, Bergquist et al. 2002). Individual aggregations may be restricted to an area of less than 1 m2, common at the sites in Garden Banks and Vioska

Knoll (Chapter 6), or can form large clusters of aggregations covering hundreds of m2 at

GC234 and GC232 (MacDonald et al. 1990, MacDonald et al. 2003). Model results suggest that the size of an aggregation is controlled mainly by the size of the substratum available for settlement (Chapter 2).

27 The environment surrounding tubeworm aggregations in this stage, less than 10-

15 years from the onset of tubeworm recruitment, is characterized by high concentrations of reduced chemicals (Bergquist et al. 2003b, Chapter 4, 5). The methane present results in elevated sulfate reduction rates by microbial consortia (Aharon and Fu 2000, Boetius et al. 2000). The continued abundance of sulfide among the tubeworm tubes apparently restricts community composition, favoring the persistence of endemic species (Bergquist et al. 2003a, Chapter 4). B. childressi remains present, along with its associated gastropod Bathynerita naticoides and the endemic gastropod Provanna sculpta. Other dominant endemic fauna include the polynoid polychaetes Harmothoe sp. nov. and

Branchinotogluma sp. nov., and the bresilid shrimp Alvinocaris stactophila. The remainder of the community is largely comprised of colonist fauna that attain greater abundance at, but are not restricted to seeps. The most common of these species are the gastropods Cancellaria rosewateri and Cataegis meroglypta, and the sipunculid

Phascolosoma turnerae (Chapter 4).

When tubeworm aggregations reach 20-30 years of age, sulfide concentrations are much reduced (Chapter 4). Declines in seepage rate result from ongoing carbonate precipitation, both at the sediment surface and below (Chapter 3). The development of a significant subsurface mass of tubeworm roots may further occlude seepage pathways. As seepage rates subside, mussel population size and tubeworm settlement rate will decline

(Bergquist et al. 2002, Chapter 4). Fewer B. childressi are present due to lower concentrations of methane in the epibenthic layer. Tubeworm recruitment is still ongoing, but the rate is reduced due to the monopolization of existing substrata, reduced rates of substrate formation through carbonate deposition at the sediment surface, and reduced

28 availability of epibenthic sulfide (Bergquist et al. 2003b, Chapter 2, 4). The existing tubeworm population will remain viable due to their increasing reliance on their roots for sulfide uptake (Chapter 3).

The rapid growth of young tubeworms, up to approximately 10 cm per year

(Bergquist et al. 2000, Chapter 5), in conjunction with reductions in seepage rate increase the amount of tube surface area devoid of sulfide (Bergquist et al. 2003a). As levels of reduced chemicals in the habitat subside (Bergquist et al. 2003b), more of the non- endemic background fauna are capable of colonizing the seep habitat (Chapter 4). This leads to a more diverse community than is found in mussel beds or very young tubeworm aggregations (Carney 1994, Bergquist et al. 2003, Bergquist et al. in press, Chapter 6).

Amphipods, chitons and limpets begin to take advantage of the remaining bacterial biomass in the aggregations (Chapter 4, 5). While these additions to the community increase the diversity of the fauna, the overall biomass of the associated community lessens likely reflecting reductions in primary productivity (Chapter 4). The onset of colonization by species in higher trophic levels, including multiple species of polychaetes such as the predatory Eunice sp. nov. that forms tubes on the anterior ends of the tubeworms, results in additional reduction of primary consumer biomass (Bergquist et al.

2003a, Chapter 4).

Further abatement of seepage occurs with continued carbonate deposition and biotic influences (Chapter 3). Extensive tubeworm root development results in rapid depletion of sulfide from pore fluids and imposes a physical barrier to seepage. The theoretical release of tubeworm symbiont-generated sulfate into deeper sediment layers may also contribute to diminishing sulfide concentrations (Chapter 3). The release of

29 sulfate supplies an energetically favorable oxidant to deeper sediment layers, thereby augmenting integrated rates of anaerobic hydrocarbon oxidation. The site of sulfide generation shifts deeper within the sediments, reducing the amount of sulfide produced near the sediment surface. Increased rates of hydrocarbon oxidation also enhance carbonate deposition, further restricting seepage pathways.

As sulfide concentrations decline, the recruitment rate declines and finally ceases

40 to 60 years after the local seepage source was formed (Bergquist et al. 2002, Chapter

2). By 50 years of age, growth rates of L. luymesi and S. jonesi slow to less than 1 and 0.5 cm per year respectively (Chapter 5). However, even some the largest tubeworms measured exhibited some growth, suggesting that there is no distinct asymptote to tubeworm size (Bergquist et al. 2000, Chapter 5). The spherical habitat generated by the tubeworms may be massive at this stage, with L. luymesi length normally ranging between 85 and 120 cm at 50 years (Chapter 5). Sulfide is occasionally still detectable in aggregations of this age, but is more highly spatially variable (Bergquist et al. 2003b,

Chapter 5).

The community inhabiting tubeworm aggregations in this stage continues to be comprised of a mixture of endemic and non-endemic species, though shifts in community structure are apparent. The endemic shrimp Alvinocaris stactophila significantly declines in abundance and shows similar distribution patterns to the endemic galatheid

Munidopsis sp. 1. These species are replaced by the shrimp Periclemenes sp. and

Munidopsis sp. 2 that significantly increase in abundance with aggregation age (Chapter

4). It has been hypothesized that the former species have higher tolerance to reduced chemicals and anoxic conditions, while the later are capable of out-competing them in

30 more benign habitat conditions (Chapter 4). Diversity in this stage remains fairly high, as the number of species in a tubeworm aggregation is directly related to the size of the habitat (Bergquist et al. 2003a, Chapter 4). The aggregation containing the greatest number of associated species (47) was in this stage, and was one of the largest aggregations collected (726 tubeworms, 4.3 m2 tube surface area, Chapter 6). Reduced availability of localized chemosynthetic production leads to continued declines in

primary consumer and overall biomass (Chapter 4). The community becomes dominated

by secondary consumers and scavengers taking advantage of the physical structure

created by the tubeworms. The majid crabs Rochinia crassa and R. tanneri, the giant isopod Bathynomus giganteus, the seastar Sclerasterias tanneri, as well as a number of

fishes including synaphobranchid eels and the hagfish Eptatretus sp. appear in older

aggregations (Carney et al. 1994, Bergquist et al. 2003, Chapter 4). The proximity of

tubeworm aggregations in earlier successional stages may allow these large, mobile

species to forage in areas of greater prey abundance without remaining in the toxic

environment for extended periods of time (Chapter 4).

This stage may persist for hundreds of years with continued slow tubeworm

growth and little or no seepage of reduced chemicals in the water column. These are the

most long-lived non-colonial animals known, based on growth data and the maximum

size of the tubeworms collected (Bergquist et al. 2000, Chapter 5). A 2.8 m L. luymesi

individual was estimated to be over 400 years of age and a 1.1 m S. jonesi individual was

estimated to be nearly 300 yrs of age (Chapter 5). The extreme longevity of these

tubeworms is related to the low mortality rates they exhibit. Mortality rate is less than

0.1% annually for L. luymesi individuals in excess of 100 cm in length and approximately

31 50-60 years of age (Chapter 2). Eventually, continued tubeworm growth leads to

“basketing” of the aggregations with the tubeworms lying recumbent on their sides along the sediment surface (Bergquist et al. 2003a). However, this does not necessarily indicate that the individual tubeworms are in some way senescent, as their physiological condition does not appear to decline in this stage (Bergquist et al. 2003b).

In the oldest tubeworm aggregations, a lack of chemosynthetic production by free-living bacteria leads to the near elimination of the lower trophic levels. However, some endemic species remain are present in lower numbers in the later successional stages (Chapter 4). The endemic species are normally scavenging or predatory species, including Munidopsis sp. 1, Harmothoe sp. nov. and Branchinotogluma sp. nov. (Chapter

4). Filter feeding organisms such as hydroids and serpulid polychaetes increase in abundance, taking advantage of the tubeworm tubes significantly elevated above the substrate to feed in higher flow regimes (Chapter 4). Members of higher trophic levels also continue to inhabit the aggregations and likely forage over the surrounding seafloor and within adjacent aggregations.

This linear progression through temporal successional stages may not be consistent in aggregations in different habitats (Chapter 6). Tubeworm aggregations at brine seeps tend to consist of a greater proportion of S. jonesi and are generally restricted to isolated patches. The persistence of B. childressi in somewhat older aggregations indicates that these brine dominated sites also contain higher methane concentrations in the epibenthic layer. Sulfide concentrations are also higher than expected in some older aggregations, leading to the persistence of the endemic heterotrophic fauna (Chapter 5).

The aggregations found in Garden Banks, the brine pool and in certain locations within

32 Mississippi Canyon and GC354 fall into this category. While the general successional trends described above (reductions in biomass, shifts in trophic structure, etc.) remain true for these sites, their trajectory through the successional sequence is somewhat altered.

Deep Seep Communities

The deeper seeps of the Gulf of Mexico are far less well explored than those

of the upper slope. General descriptions of the geological setting and the

chemosynthetic fauna inhabiting the seeps exist (Paull et al. 1984, Cary et al. 1989,

Brooks et al. 1990, Bryant et al. 1990, MacDonald et al. 2003). However, there is a

paucity of information on the biodiversity and community structure of these seeps,

the exception being of the mussel beds at Florida Escarpment (Turnipseed et al. 2003,

VanDover et al. 2003, Chapter 6).

Florida Escarpment mussel beds are comprised of Bathymodiolus brooksi and

B. heckeri. They are suggested to be the most diverse of all vent and seep mussel bed

communities. Turnipseed et al. (2003) used species rarefaction curves to find an

expected value of approximately 55 species (retained on a 250 µm sieve) to be

identified in a collection of 10,000 individuals. Mussel beds are dominated by the

ophiuroid Ophiurocten spinilimbatum along with the gastropod Fucaria sp. and the

limpet Paraleptopsis sp. The bresilid shrimp Alvinocaris muricola and the polynoid

Branchinotogluma sp. are also common inhabitants of the mussel beds (Chapter 6).

33 Escarpia laminata is the only known vestimentiferan from the Florida

Escarpment. This tubeworm co-occurs with the chemoautotrophic mussel B. brooksi,

though the sulfide sources of the two appear to be different (Cary et al. 1989).

According to Cary et al. (1989) E. laminata has a sulfur stable isotope signature

consistent with a biogenic sulfide source, while B. brooksi utilizes sulfide generated

within the Florida platform. It appears that the tubeworm aggregations at this site may

be more diverse than the mussel bed communities, similar to the situation of the ULS.

In a recent study, there were 21 species identified among the 256 individuals (retained

on a 2 mm sieve) of associated fauna in a single tubeworm community collection and

19 species identified among the 1898 individuals collected in 2 mussel bed

collections (Chapter 6). While diversity was higher in the tubeworm aggregations,

abundance and biomass (per unit area) of associated fauna were lower, again

consistent with our current knowledge derived from the ULS seeps. Most of the

dominant fauna were consistent between tubeworm aggregations and mussel beds,

with the addition of a species of the galatheid Munidopsis and 3 unidentified species

of amphipods.

The Atwater Valley site in AT425 contains areas of seepage supporting B. brooksi and E. laminata populations and individuals of an undescribed species of lamellibrachid.

Only one community collection has been obtained from a single mussel bed at this site.

This collection consisted solely of Alvinocaris muricola and Ophiurocten spinilimbatum

(Chapter 6). Other species that have been noted at the site are members of the background

34 fauna including the holothurian Benthodytes lingua, sessile octocorals and a stalked anemone (MacDonald et al. 2003).

At the Alaminos Canyon site, extensive beds of B. brooksi and B. childressi and tubeworm aggregations of Escarpia laminata are present (Carney et al. 1994, Chapter 6).

A species of Lamellibrachia has also been reported (Nelson and Fisher 2000), though its taxonomy and relationship to the species at Atwater Valley remain unresolved. An unidentified species of lucinid and a thyasirid have also been collected at this site

(Chapter 6). The heterotrophic community is again dominated by Alvinocaris muricola and Ophiurocten spinilimbatum (Chapter 6). A. muricola appears here in much higher abundance than at either of the other two deep seep sites, with O. spinilimbatum showing a concomitant decline in abundance. Other associated fauna include a sipunculan

(Phascolosoma sp.), a galatheid (Munidopsis sp. different from the upper slope species), and a species of limpet (Carney 1994, Chapter 6). Like the other sites examined, diversity of associated fauna appears to be higher in tubeworm aggregations than mussel beds.

This is apparent in measures of species richness with 19 species found in 2 tubeworm collections, and 12 species found in 2 mussel beds, and in the Shannon-Weaver diversity index with 1.20 and 0.57 averages for tubeworm and mussel communities respectively.

Conclusions

The seeps of the Gulf of Mexico have been studied extensively over the past 20 years. The knowledge gathered from these investigations has changed our perception of the ecology of cold seeps in general. Our understanding of seep tubeworm nutrient

35 acquisition, population dynamics, and phylogeography have benefited greatly from these studies. Some of the greatest progress has been made in the recent studies of community ecology in the GoM. On the upper slope, there now exists a successional model describing the shift in community structure from mussel beds through tubeworm aggregations of increasing age. There are far fewer examinations of the ecology of the deeper seep sites in the GoM, though some of the trends observed in mussel bed and tubeworm dominated communities from the ULS hold for these habitats as well. It remains to be seen how well the patterns shown in the GoM translate to the other seeps around the world. It is our hope that this review provides the impetus not only for additional exploration of the global seep environment, but for the testing in other habitats of the hypotheses generated from these investigations of Gulf seep ecology.

Chapter 2

High Hydrogen Sulfide Demand of Tubeworm Aggregations Modifies the Chemical Environment at Deep-Sea Hydrocarbon Seeps

Abstract

Lamellibrachia luymesi is a long-lived vestimentiferan polychaete that produces biogenic habitat at hydrocarbon seeps on the upper Louisiana slope of the Gulf of

Mexico. L. luymesi relies on endosymbiotic, chemoautotrophic bacteria for nutrition which are supplied with hydrogen sulfide acquired from seep sediments by the tube worms. In this study, an individual-based model is developed for L. luymesi aggregations.

Results show that aggregations can persist for centuries due to extremely low mortality rates. Recruitment patterns reflect intra-specific competition for settlement space, with the recruitment period estimated to be between 11 and 68 years. Substantial hydrogen sulfide requirements are estimated for large aggregations of L. luymesi, exceeding 30 mmol·hr-1. In addition to modifying habitat through physical structure, L. luymesi may be

considered an ecosystem engineer because of its profound effect on the chemical

environment at hydrocarbon seep sites.

Introduction

The ability of an organism to create and modify habitat through the alteration of

physical structure and the chemical environment, ecosystem engineering, has become

37 recognized as an important factor influencing community structure in many systems

(Jones et al. 1994). In hydrothermal vent and hydrocarbon seep ecosystems, vestimentiferan tube worms (Polychaeta: Siboglinidae) are often the dominant organisms in terms of biomass and abundance (MacDonald et al. 1990, Tunnicliffe 1991, Lutz &

Kennish 1993). At hydrocarbon seep sites on the upper Louisiana slope of the Gulf of

Mexico, the vestimentiferan Lamellibrachia luymesi forms aggregations of hundreds to thousands of individuals, creating large complex structures in an otherwise barren landscape (Kennicutt et al. 1985, Carney 1994). This physical structure provides biogenic habitat for a large community of organisms, including 18 known endemic species

(Berquist et al. 2003a).

In order to understand the production and maintenance of biogenic habitat, knowledge of the population dynamics of the organisms that create it is essential.

Recently, a theoretical model of L. luymesi aggregation development and community succession was proposed by Bergquist et al. (2003a). This model suggests that aggregations become established on hard substrata in the presence of active hydrogen sulfide (H2S) seepage. Communities associated with these nascent aggregations are

mainly comprised of endemic species with high sulfide tolerances. The period of vestimentiferan recruitment was calculated to persist on the order of a few decades until

surface expression of sulfide diminishes and additional substrate is unavailable

(Bergquist et al. 2002). At this time, non-endemic species begin to utilize the biogenic habitat of L. luymesi aggregations where they were previously excluded because of their lower tolerance to environmental sulfide. This stage of community succession may persist for centuries, as the maximum life span of L. luymesi has been estimated to exceed

38 250 years (Fisher et al. 1997, Bergquist et al. 2000). While the ecological consequences of diminishing sulfide expression have been investigated, the mechanisms generating the observed decline in seepage rates over time are poorly understood.

In addition to producing biogenic habitat, it is possible that L. luymesi acts as an ecosystem engineer by altering the chemical environment of these hydrocarbon seep sites. All vestimentiferans rely entirely on sulfide-oxidizing bacterial endosymbionts for their nutritional requirements (Childress & Fisher 1992, Nelson & Fisher 1995). L. luymesi differs from most species of vestimentiferans in that it is capable of using posterior extensions of its body and tube, the “root”, to acquire H2S from sediments

(Julian et al. 1999, Freytag et al. 2001). It is possible that increasing population-level uptake over the course of aggregation development could deplete sulfide resources in hydrocarbon seep sediments. To determine if this is plausible, estimates of sulfide uptake rates for intact aggregations of different ages are required. The collection of these data in

situ is impractical if not impossible due to the logistical constraints of sampling with submersibles in the deep-sea and the extreme longevity of L. luymesi. Therefore,

ecological modeling techniques are the best approach to obtaining this information.

The application of individual-based models (IBMs) to modeling the long-term

dynamics of populations has greatly increased in the last 10-15 years (Grimm 1999). One

of the benefits of IBMs is that they are capable of reproducing the variability of natural systems at a relatively low level of model complexity (Schmitz 2000, Pascual et al.

2002). These models have the additional advantage of producing results in the form of

testable hypotheses (Fahse et al. 1998, Newton et al. 2001). IBMs are particularly useful

in ecological investigations of long-lived species where the data required to generate

39 more classical population-level models are difficult to collect due to the time scale involved (Hunter et al. 2000).

In this study, an individual-based model is developed to address previously intractable questions pertaining to L. luymesi ecology on the upper Louisiana slope.

Mortality and recruitment rates are estimated from empirical data. The model is validated by comparing model output to observed population characteristics. Once the validity of the model is established, population-level sulfide uptake rates are estimated. The goal of this study is to examine the long-term population dynamics and sulfide uptake of L. luymesi aggregations in order to better understand the degree to which these aggregations are capable of altering hydrocarbon seep habitats through the persistence of physical structure and modification of the chemical environment.

Methods

Lamellibrachia luymesi aggregations were collected in 1997 and 1998 using the

Johnson Sea-Link submersible. Whole aggregations were collected using a hydraulically activated collection device lined with a 64µm mesh net (see Bergquist et al. 2002 for description). A total of 12 aggregations were haphazardly collected from two sites on the upper Louisiana slope: Green Canyon 234 located at 27°44.75’N; 91°13.31’W between

540 to 560 meters depth, and Bush Hill located at 27°46.96’N; 91°30.46’W at 540 meters depth.

The length of each tubeworm tube in each of the twelve aggregations was measured to the nearest 1mm. Growth rates were estimated by measuring a subset of

40 individuals which had been previously stained with a chitin stain, as presented in

Bergquist et al. (2000). Size-specific growth rate was determined by non-linear regression (Bergquist et al. 2000). Linear regression of tube length to the residuals of the growth model was used to estimate the error associated with growth rate. The model contains a stochastic error term varying within the prediction limits (3 standard deviations) of this linear regression. All regression analyses were carried out using the statistics package JMP® (SAS Institute, 1997). With the inclusion of the error term, the growth equation is as follows:

−0.0103Lt Lt + 1 = Lt + ()3.884 ⋅ e + (ε(N(0,1))⋅3 1.775 − 0.00474Lt ) (1) where Lt = length of individual at time t, and ε(N(0,1)) is a normally distributed random deviate with mean of 0 and standard deviation of 1.

Mortality rates were approximated by the frequency of empty tubes found in the twelve sampled populations. Population-wide average mortality rate was calculated as the number of empty tubes divided by the total number of individuals (including empty tubes) in the collected aggregations. For model implementation, size-specific mortality rates were estimated by non-linear least-squares regression of size class to the frequency of empty tubes using the following function:

-ωL pm = υe t (2)

where pm = mortality probability, and υ and ω are constants.

Recruitment rate was modeled as a population-level process. This was necessary as data on individual reproductive output of L. luymesi, or any other seep vestimentiferan, are lacking. The approach described here is a modification of classical estimations of

41 initial cohort size from static life-table data (e.g. Lowe 1969). The number of individuals of a certain 10 cm size class at a given time is a function of the initial number of recruits, mortality rate, and growth rate. This was simulated by iteration of a separate individual- based model. Recruitment classes were followed through time until at least 10 recruitment events resulted in a cohort with the same average length and number of individuals as the observed size class. These data were fit to the total number of individuals in the population at that time using a modified stock-recruitment model

(Maynard Smith & Slatkin 1972, Shepherd 1982, Shea et al. in prep.):

aNt R = (3) aNt c 1+ b()K where R = number of recruits, Nt = population size at time t, K = carrying capacity, and a, b, and c are constants. Carrying capacity (K) was estimated as the number of individuals in the population at the time of collection for aggregations that did not exhibit active recruitment. The value of a describes the rate of population growth in the absence of density dependence, b is a relative measure of the degree of density dependence, and c is a shape parameter. During model implementation, the value of a was modified at each time step by a normally distributed stochastic term with a mean of 0 and standard deviation derived from the pooled sum of squared error (SSE) associated with the estimates of a (0.0872). When population size equaled or exceeded carrying capacity, recruitment rate was set to zero. Recruitment period was estimated as the period of time between the start of a model realization and the end of recruitment.

To determine an individual tube worm’s mass-specific sulfide uptake rate, length

1.388 was converted to mass by the function Mi = 0.0088⋅ Li as modified from Bergquist

42

(2002) where Mi = soft tissue wet weight excluding tube material in grams, and Li = tube length in cm. Sulfide uptake rates measured for L. luymesi in the laboratory were between

1.60 and 5.98 µmol·g-1·hr-1 (Freytag et al. 2001). The minimum uptake measured was used as the basal requirement. Total individual uptake was determined by the sum of this basal requirement and that required for growth, where the highest annual growth rate

(10cm·yr-1) reflected maximum measured uptake (5.98 µmol·g-1·hr-1):

⎡ ⎤ ⎢ ⎛ gi ⎞⎥ Ui = Mi⎢1.60 + ⎜4.38⋅ ⎟⎥ (4) ⎢ ⎥ ⎣ ⎝ 10 ⎠⎦

-1 where Ui = individual uptake rate in µmol·hr , Mi = mass in g, and gi = individual growth rate in cm as determined for that time step. Aggregation uptake rate was determined by summing uptake across all individuals alive in the population at the end of each time step.

Two sets of realizations of the population growth model were carried out to investigate different aspects of this study. Initially, recruitment rates were entered into the model by setting the values of the parameters in equation (3) according to the aggregation-specific values. Duration of recruitment period was estimated as the time when population size first reached K. Validity of the model was evaluated for each aggregation by comparison of generated size-frequency distributions to actual size- frequency using a log-likelihood test (Zar 1984). Once this analysis was conducted, the sulfide uptake function in the form of equation (4) and a general recruitment rate were entered into the model. Recruitment rate was determined at the start of each realization by selecting values of a, b, and c in equation (3) from the distribution of values estimated.

A normal distribution was used to select values of a (mean ± sd, 0.600 ± 0.199, n=9) and c (mean ± sd, 5.34 ± 1.59, n=9) and log-transformed values of b (mean ± sd, 6.88 ± 2.53,

43 n=9). Carrying capacity was set at 1000 individuals during these simulations, the approximate size of the largest aggregations collected. The rate and variability of sulfide uptake for generalized L. luymesi aggregations was then determined.

Results

Mortality rates of L. luymesi are extremely low. A total of 32 empty tubes were found out of the 4618 individuals collected. Assuming that all empty tubes represent worms which died in the last year, the average mortality rate is 0.7% per year. Mortality is highest in juvenile tube worms, with the model predicting only 2.5% mortality per year for the smallest individuals (Fig. 2-1). Mortality quickly decreases as tube worms increase in length. For individuals measuring 31.5 cm, 1% mortality is predicted. This extremely low mortality rate is reflected in the original data in that only one empty tube was recorded for all individuals over 70 cm collected (n = 1498).

44

0.035

0.03

0.025 e 0.02 empirical model 0.015 95% CI mortality rat mortality 0.01

0.005

0 0 20 40 60 80 100 120 140 160 180 200 size class (cm)

Figure 2-1: Population growth characteristics of Lamellibrachia luymesi:

Mortality rate. Points on the figure represent the frequency of empty tubes in each size

- class of tubeworm found in the collected aggregations. The curve is in the form pm = υe

ωL t where pm = mortality probability, Lt = length of individual tubeworm, υ = 0.244, and

ω = 0.0279.

Recruitment rates exhibited a high degree of density dependence (Fig. 2-2). All non-linear regressions of recruitment to population size were highly significant ( Table 2-

1). Average recruitment period was estimated to be between 22.4 and 41.1 years

(Table 2-2) with the range of individual simulations falling between 11 and 68 years

(Fig. 2-3). Variability among aggregations was relatively small given the large range of carrying capacities (mean ± sd for all simulations, 28.6 ± 8.42, n = 900).

45

100

80

empirical 60 mo del 95% CI 40 number of recruits of number 20

0 0 200 400 600 800 1000 population size

Figure 2-2: Population growth characteristics of Lamellibrachia luymesi: Recruitment rate of aggregation BH7. Numbers of recruits entering the population at a given population size were estimated from size-frequency data for the 9 aggregations which did not exhibit active recruitment (see text for details). Fitted to these data was the function aNt R = where R = annual number of recruits, Nt = number of individuals in the aNt c 1+ b()K population, a = 0.304, b = 15369, c = 5.55, and K = 917.

46

1200

1000

e 800

600

population siz 400

200

0 0 20 40 60 80 100 120 140 160 180 200 time (years)

Figure 2-3: Model output for Lamellibrachia luymesi aggregation BH7. Presented are annual means and 95% confidence intervals of population size for 100 realizations of 100 years.

47

Table 2-1: Results of non-linear regression of number of Lamellibrachia luymesi recruits to aNt population size. The nonlinear regression function is as follows: R = where R = aNt c 1+ b()K annual number of recruits, and Nt = number of individuals in the population at time t. . See Figure 2-2 for example. Only those aggregations which did not exhibit active recruitment were used in the analysis. Naming scheme follows Bergquist et al. (2002) where “BH” stands for Bush Hill site and “GC” stands for Green Canyon site (see methods for details). aggregation K a b c F P

BH3 81 0.862 169 5.68 248.9 <0.0001

BH4 458 0.743 113 3.8 22.5 0.0003

BH5 86 0.708 220 4.82 86.8 <0.0001

BH6 101 0.407 136689 8.65 34.5 <0.0001

BH7 917 0.304 15369 5.55 565.9 <0.0001

GC1 107 0.504 616 4.51 66.9 <0.0001

GC2 51 0.495 5032 6.9 33.3 0.0001

GC5 245 0.856 123 4.1 222 <0.0001

GC7 134 0.524 239 4.05 19.3 0.001

Average size frequency histograms generated from 100 iterations of the model matched the empirical data relatively well. A log-likelihood test indicated significant differences in the size distributions of 5 of the 9 aggregations examined (Table 2-2).

However, those that were quantitatively different were qualitatively similar with respect to the distribution of individuals among size classes ( Fig. 2-4 ). The primary difference in the shape of the size-frequency histograms was the tendency of sampled populations to skew to the right while model populations skewed to the left.

48 Table 2-2: Aggregation-specific recruitment period and comparison of model output to sampled size-frequency of Lamellibrachia luymesi. Estimates of recruitment period (defined as the period of time from the start of recruitment until population size is equal to or greater than carrying capacity) and size-frequency distributions were generated by 100 realizations of the population growth model using aggregation-specific parameters as listed in Table 1. Asterisks denote significant differences between simulated and actual size-frequency distributions. See Figure 4 for example. Naming scheme follows Bergquist et al. (2002) where “BH” stands for Bush Hill site and “GC” stands for Green Canyon site (see methods for details).

recruitment period size-frequency log-likelihood ratio

aggregation mean (sd) G P v

BH3 28.5 (5.68) 21.3 0.006* 8

BH4 28.2 (2.66) 82 <0.001* 10

BH5 25.5 (5.05) 22 0.015* 10

BH6 26.9 (10.65) 0.091 >0.999 11

BH7 41.1 (6.79) 112.5 <0.001* 39

GC1 26.7 (4.62) 11.3 0.586 13

GC2 22.4 (8.39) 9.27 0.597 11

GC5 35.9 (3.60) 2.35 >0.999 18

GC7 22.4 (3.16) 25.1 0.003* 9

49

300

250 s 200

empirical 150 model

100 number of individual of number

50

0 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 size class (cm)

Figure 2-4: Size-frequency of Lamellibrachia luymesi in aggregation BH7. Grey bars are averages for 100 simulated populations. Error bars represent standard deviations. Black bars are actual frequencies of individuals in the aggregation. The actual and model distributions of this aggregation were significantly different as determined by the log- likelihood test statistic (G = 112.5, P < 0.001).

Sulfide uptake increases over the lifespan of L. luymesi. While very low rates were estimated for small individuals, uptake for large tube worms often exceeds 30

µmol·hr-1, and may be as high as 50 µmol·hr-1. In whole aggregations, low rates of uptake

(<1 mmol·hr-1) are maintained for approximately 20 years ( Fig. 2-5 ). At 100 years, the average uptake rate for an aggregation is 18.1 mmol·hr-1 (SD = 0.694 mmol·hr-1). In older aggregations, uptake rate continues to increase as the individuals in the population increase in size. In a 200 year old aggregation consisting of 1000 individuals, sulfide uptake is approximately 30 mmol·hr-1(mean ± sd, 30.26 ± 0.919 mmol·hr-1).

50

35

30

25

20

15

10 sulfide uptake (mmol/hr)

5

0 0 20 40 60 80 100 120 140 160 180 200 time (years)

Figure 2-5: Hydrogen sulfide uptake rate for whole Lamellibrachia luymesi aggregations. Presented are means and 95% confidence intervals for 100 realized aggregations with a carrying capacity of 1000 individuals.

Discussion

The sulfide demand predicted in this study provides a mechanism for L. luymesi to alter its environment not only by producing persistent physical structure, but also by mediating the abundance of sulfide in their habitat. By using a modeling approach to address questions of in situ sulfide demand, we set up a framework within which we could evaluate the impact of L. luymesi on hydrocarbon seep habitats at temporal scales beyond the scope of a normal sampling program. In addition, the estimation of sulfide uptake rates for intact aggregations would have been impossible to obtain with a submersible using existing sampling methods.

51 In order to develop this model, a number of simplifications were necessary. The most significant of these was that the aggregations were not simply made up of one species. There are two, and rarely three species of vestimentiferans at these sites

(McMullin et al. 2003). The newly described Seepiophila jonesi (Gardiner et al. 2001) is normally present in these aggregations, though less abundant than L. luymesi (Bergquist et al. 2002). The third species appears to be a distinct species within the escarpid group

(including the genera Escarpia, Paraescarpia, and Seepiophila) based on mitochondrial

COI sequence data (McMullin et al. 2003). However, this species has only rarely been collected and is not yet described. This suggests that competition may be affecting the dynamics modeled here. Unfortunately, little empirical data exist on S. jonesi, and no data exist on the third species, precluding their inclusion in the model. Future investigations are required to elucidate their role in the communities at these sites.

In the model, empty tubes were assumed to represent individuals that died within one year of collection. This was a conservative assumption based on the finding that the less robust tubes of the hydrothermal vent vestimentiferan Riftia pachyptila required one year to fully dissolve (Gaill et al. 1996). Preliminary investigations indicate that the dissolution rates for tubes of cold seep species are four times slower (F. Gaill, pers. comm.). Because there are few or no natural predators of L. luymesi (Carney 1994,

MacAvoy et al. 2002), removal of empty tubes or living individuals is very unlikely, especially considering that the root extends into the sediment 1 to 2 times the length of the exposed anterior portion (Freytag et al. 2001). Therefore, the annual mortality rates presented here are maximum estimates of actual mortality in situ.

Modeled populations approximated the empirical data very well. The stock-

52 recruitment function tended to overestimate recruitment rate at the largest population sizes, leading to rapid cessation of recruitment. Despite this incongruity, the smallest size classes are still adequately represented in modeled populations. However, the few individuals in the largest size classes of sampled aggregations were not produced by the model. Disparities between real and modeled populations could result from natural growth rate outliers or slower initial recruitment rates. Since space could not be adequately parameterized and included in the model, small scale spatial effects may also contribute to these differences (Pascual et al. 2002).

It is hypothesized that density-dependence in the form of intra-specific competition is responsible for the recruitment patterns observed. Initial increases in recruitment rate are believed to be related to changes in substrate availability rather than increased local larval supply because of the similarity in genetic variability within and between L. luymesi aggregations in the Gulf of Mexico (McMullin 2003). Authigenic carbonate deposition, occurring as a by-product of coupled sulfate reduction and methane oxidation (Kennicutt et al. 1989, Boetius et al. 2000), or exposure of carbonates due to sediment slumping and salt tectonism (Ferrell & Aharon 1994, Roberts & Aharon 1994) could provide substrate for primary succession. While gregarious settlement on tubes of living worms often occurs in hydrothermal vent vestimentiferans (Fisher et al. 1990) this is quite rare in Gulf of Mexico seep species. Empty tubes of dead vestimentiferans also occasionally provide substrate with individuals settling on the interior and growing their roots down existing tubes (Bergquist et al. 2002). Increased availability of carbonate and gregarious settlement would ameliorate intra-specific competition for space leading to increasing recruitment rates in nascent aggregations.

53 In older aggregations, enhanced intra-specific competition for space is the likely mechanism generating the apparent density-dependent reduction in recruitment found here. The decline of sulfide expression prevents both recruitment and carbonate deposition since they occur only in the presence of elevated concentrations of sulfide and methane at the sediment surface (Roberts & Aharon 1994, Fisher et al. 1997). In addition, the accumulation of sediment around older aggregations could bury carbonate substrata.

While a very slow process in the Gulf of Mexico (Aharon & Fu 2000), sedimentation is likely enhanced by the baffling effect of aggregations (Fisher et al. 1997). In this way, aggregations could alter their habitat by reducing the amount of exposed carbonate, preventing additional recruitment.

In established aggregations, population size is stable over time periods measured in centuries. This is due to the extremely low mortality rates of large individuals and the cessation of recruitment. The occasional mortality event would be offset by the concomitant provision of substrate for gregarious recruitment, further stabilizing population size. Populations exhibiting long term stability are expected for organisms with high population growth rates and overlapping generations, traits most commonly exhibited by vertebrate populations (Lindström & Kokko 2002). Stability is further enhanced through the cohort effect if populations are composed of highly variable individuals (Lindström & Kokko 2002), emphasizing the relative importance of long- distance dispersal in maintaining variability in L. luymesi populations. Although existing growth data indicate that individual tube worms can live for over 250 years (Fisher et al.

1997, Bergquist et al. 2000), an upper limit on longevity of individuals or aggregations could not be estimated due to the paucity of mortality data for tube worms over 2m.

54 Using the average mortality rate (0.7% per year) and the inverse relationship between mortality rate and longevity (Case 2000), the average life span of L. luymesi is estimated to be 144 years. Considering that the majority of observed mortality was in young individuals, this average life span could only be produced if much of the population lived for over 200 years. In addition, constraints imposed by the maximum size of sampling gear prevented the collection of larger aggregations, including uninterrupted fields of tube worms covering hundreds of square meters, which likely persist far longer than the aggregations serving as the basis for the model. All available data support the hypothesis that the physical structure produced by any single aggregation persists for at least 200 years.

Estimated sulfide demand of large, old aggregations exceeds 30 mmol·hr-1.

Certain characteristics of the laboratory study on which the individual uptake rates were based (Freytag et al. 2001) suggest that this represents an overestimate of in situ uptake.

The individuals used in the study were all less than 50 cm in length (John Freytag, pers. comm.). Given the well demonstrated allometric scaling of metabolism (Schmidt-Nielsen

1975), larger individuals should have lower mass-specific sulfide demand than small tube worms. This allometric effect was approximated by scaling uptake with growth rate, which declines steadily with length (Bergquist et al. 2000). The highest uptake rates (5.98

µmol·g-1·hr-1) were assigned to the fastest growing (smallest) individuals, and the lowest rates (1.60 µmol·g-1·hr-1) to the slowest growing (largest) individuals. In this way, massspecific rates are effectively scaled over a factor of 3.7. This is similar to the factor of 3.6 one would expect over the range of masses in this study (1 to 14 g) calculated from the universal relationship between mass and respiration rate, where mass-specific

55 respiration rate = 3.8 * mass-0.25 (Schmidt-Neilsen 1975). It remains possible that rates derived from the small size range of tube worms used in the Freytag et al. (2001) study do not represent the full range of uptake rates in situ, and may overestimate uptake for the largest individuals. Thus the results of this study should be viewed as a testable hypothesis (Newton et al. 2001) with the absolute value changing as parameter estimates are refined.

Intra-specific competition could reduce natural sulfide uptake rates through local depletion or direct root contact. The effective surface area available for sulfide uptake would be reduced by contact between adjacent roots. While the upper portions of roots are often in contact with one another, little information exists on the overall spatial arrangement of root masses despite ongoing efforts to visualize and collect them intact.

Local depletion of sulfide by neighboring individuals could affect in situ rates due to the positive correlation between sulfide concentration and uptake rate (Girguis et al. 2002).

However, the sulfide concentrations used in the laboratory study, 218 to 431 µmol·l-1

(Freytag et al. 2001), were well below the highest concentrations measured in sediments around the periphery of aggregations, 12.7 mmol·l-1 (Arvidson et al. 2004). Local depletion effects may be more significant in older aggregations if subsurface sulfide concentration declines over time. The paucity of data on temporal changes of sulfide concentration beneath aggregations prevents explicit modeling of this parameter. If roots are densely aggregated and subsurface sulfide concentration is limiting, intra-specific competition will likely affect uptake rates in natural populations.

The long term persistence of L. luymesi aggregations in conjunction with their apparent high sulfide demand suggest that a mechanism exists for maintaining elevated

56 concentrations of available sulfide. The known sources of sulfide in seep sediments are reduction of seawater sulfate and flux from various deep sources. While sulfate reduction likely accounts for a large proportion of the available sulfide due to the extremely high reduction rates measured at upper Louisiana slope seep sites (Aharon & Fu 2000), seawater sulfate is limited to the upper 20 to 30cm of sediment (Arvidson et al. 2004).

It has been suggested that L. luymesi could augment sulfide production in deeper sediments by releasing sulfate through their roots (Julian et al. 1999, Freytag et al. 2001).

Sulfate reduction around roots would allow for authigenic carbonate deposition at depth in seep sediments. Carbonate formation along with growth of roots through existing seepage conduits could effectively cap the hydrocarbon seep, trapping and pooling sulfide beneath tube worm aggregations. It is possible that increased sulfate supply along with pooling of sulfide could maintain the elevated nutrient levels necessary for the long term persistence of aggregations. Future investigations could determine if known sources are sufficient to maintain sulfide at steady state concentrations in the presence of the high uptake rates of L. luymesi aggregations.

In conjunction, these processes would have a considerable impact on the distribution of chemicals in the water column around aggregations as well as in hydrocarbon seep sediments. In addition to pooling sulfide, the capping effect of aggregations would reduce surface expression of sulfide at seeps (Sassen et al. 1993,

Julian et al. 1999, Freytag et al. 2001). Once water column sulfide concentrations are reduced, the biogenic habitat created by L. luymesi may be colonized by the non-endemic

“background” fauna of the upper Louisiana slope. These organisms are intolerant of elevated sulfide levels, as it is normally poisonous even in small amounts (Grieshaber &

57 Völkel 1998). In addition to the direct provision of physical structure, these mechanisms allow L. luymesi aggregations to have a profound effect on the chemical milieu of hydrocarbon seep habitats.

In summary, Lamellibrachia luymesi aggregations exhibit long-term stability and persistence produced by low mortality rates and depressed recruitment, allowing their physical structure to exist for centuries. This suggests that there is a persistent mechanism generating sulfide at sufficient rates to support the hypothesized high rate of uptake of aggregations. In addition to the provision of biogenic habitat, L. luymesi acts as an ecosystem engineer by altering the flow and distribution of sulfide and sulfate at hydrocarbon seeps.

Notes

Kat Shea, Derk Bergquist, and Chuck Fisher were coauthors on the publication of this study and assisted in the editing of the manuscript. Kat Shea assisted with the construction of the population growth model. Derk Bergquist provided data on tubeworm aggregation characteristics from previous studies. Chuck Fisher provided the inspiration for the undertaking of this project and assisted with the parameterization of the model.

Ben Predmore, Meredith Redding, Julie Barsic, Brian Tiegs, and Nicole Iacchei all contributed to the enumeration and measurement of tubeworms.

Chapter 3

Modeling the Mutualistic Interactions Between Tubeworms and Microbial Consortia

Abstract

The deep-sea vestimentiferan tubeworm Lamellibrachia luymesi is the longest- lived animal known, reaching ages in excess of 250 years. Here, we present the results of a diagenetic model supporting the hypothesis that the persistence of this tubeworm is achieved through augmentation of the supply of sulfate to hydrocarbon seep sediments.

In the model, L. luymesi releases the sulfate generated by their internal, chemoautotrophic, sulfide-oxidizing symbionts through posterior root-like extensions of their body. The sulfate fuels anaerobic methane oxidation and hydrocarbon degradation by bacterial/archaeal consortia generating sufficient sulfide to support moderate sized aggregations of L. luymesi for hundreds of years. This study expands our concept of the benefits derived from complex interspecific relationships, in this case involving members of all 3 domains of life.

Introduction

Complex positive species interactions have been shown to expand the ecological niche and increase the persistence of the organisms involved in a variety of systems. In terrestrial systems, increased diversity of mycorrhizal symbionts is correlated with increased biodiversity of plant communities, resulting in greater stability and longer

59 persistence at the community level (van der Heijen et al. 1998). Off the Atlantic coast of

North Carolina, the coral Oculina arbuscula harbors a majid crab, Mithrax forceps, which prevents overgrowth of macroalgae and shading of the corals (Stachowicz and Hay

1999). This allows Oculina to maintain its facultative mutualism with photosynthetic zooxanthellae in well-lit habitats, increasing the amount of energy available to the coral for growth and reproduction. At cold seeps in the Cascadia (Treude et al. 2003) and

Aleutian (Wallman et al. 1997) subduction zones, bioirrigation through burrow formation and bioturbation by clams (Calyptogena spp.) has been shown to significantly affect the distribution of anaerobic methane oxidation.

Lamellibrachia luymesi inhabits hydrocarbon seeps on the upper Louisiana slope

(ULS) of the Gulf of Mexico from 400-1000 m depth. L. luymesi does not posses a digestive system; rather it acquires energy via internal sulfide-oxidizing bacterial symbionts (Childress and Fisher 1992). L. luymesi differs from most vestimentiferan tubeworms by its ability to use a posterior extension of its body, the “root”, to acquire sulfide from interstitial pools in sediments (Julian et al. 1999, Freytag et al. 2001). While sulfide concentrations near the anterior plumes of the tubeworms rapidly decline below

0.1 µM as the tubeworms approach 1 m in length (Bergquist et al. 2003a), its roots allow

L. luymesi to delve into deeper sediment layers providing access to more persistent sulfide sources. In the apparent absence of lethal predation (Carney 1994, Bergquist et al.

2003b), the most significant hazard that this vestimentiferan tubeworm faces is sulfide limitation. Its high uptake rate of sulfide from hydrocarbon seep sediments, estimated at over 30 µmol·hr-1 for a moderate sized individual (Chapter 2), suggests that sulfide flux may be limiting in L. luymesi’s habitat.

60 A diverse chemosynthetic community relies on the sulfide generated as a by- product of anaerobic degradative processes in the Gulf of Mexico (Carney 1994,

Bergquist et al. 2003b). Anaerobic methane oxidation is carried out by microbial consortia consisting of sulfate reducing bacteria along with methanogenic archaea executing reverse methanogenesis (Boetius et al. 2000). Anaerobic methane oxidation and subsequent authigenic carbonate precipitation in sediments overlying gas hydrates constrains ocean-atmosphere carbon fluxes (Aliosi et al. 2002, Zhang et al. 2002), accounting for up to 20% of the global methane flux to the atmosphere (Thiel et al.

2001). Anaerobic biodegradation of large chain hydrocarbons using sulfate may be especially significant in locations such as the upper Louisiana slope where formation waters come into contact with evaporites (Röling et al. 2003) potentially altering large volumes of existing reservoirs (Aitken et al. 2004).

In this study, we address the question of whether known biogeochemical processes could supply sulfide at rates sufficient to match the requirements of long lived

L. luymesi aggregations. To augment exogenous sulfide production, it has been suggested that L. luymesi utilizes its roots to release the sulfate generated by its symbionts during sulfide oxidation (Julian et al. 1999, Freytag et al. 2001, Chapter 2). This would provide sulfate for anaerobic methane oxidation and hydrocarbon degradation at sediment depths normally devoid of energetically favorable oxidants. In a diagenetic model, the hypothesized release of sulfate is the only mechanism by which aggregations of L. luymesi can acquire enough hydrogen sulfide to persist for over 80 years. The syntophy formed among symbiotic tubeworms and microbial consortia would expand our current

61 concept for the potential complexity of positive interspecific interactions and the benefits they confer.

Methods

This study couples an individual-based population growth and sulfide uptake model (Chapter 2) to a diagenetic diffusion/advection model to compare the relative magnitude of sulfide supply and uptake for long-lived tubeworm aggregations. A series of 1000 iterations of the model under 3 different initial conditions (known sources, known sources plus root sulfate supply, and known sources with elevated seepage rates) were carried out. The rhizosphere (volume of sediment encompassed by the root system of an aggregation) is modeled as an inverted dome beneath the sediment with a radius equal to the average root length of the population (Fig. 3-1). The rhizosphere was approximated by a series of 2-dimensional discs at 2 cm intervals in order to reduce the

2- complexity of a 3-dimensional solution for a sphere of changing size. Sulfate (SO4 ),

- - + methane (CH4), sulfide (HS ), bicarbonate (HCO3 ), and hydrogen ion (H ) fluxes across the rhizosphere boundary are determined. Sulfate reduction rates using methane or larger chain hydrocarbons as electron donors are modeled in order to estimate the sulfide available to tubeworm aggregations as they change in size over the course of 250 years

62

Fig. 3-1: Model construction. Population model includes individual size-specific growth and mortality rates, and population size-specific recruitment rate. Growth rate was determined by in situ staining of tubeworm aggregations using a blue chitin stain (in situ photograph of stained aggregation demonstrating annual growth shown here) and collection after 12-14 months. Diagenetic model included advection and diffusion of sulfate, sulfide, methane, bicarbonate, and hydrogen ions as well as organic carbon content of sediments. Fluxes across the rhizosphere (root system) boundary were compared to sulfide uptake rates for simulated aggregations to determine whether sulfide supply could match the required uptake rates of aggregations.

Population growth model

The population growth model follows the methodology presented in Chapter 1 and includes population growth, mortality rate, individual growth rate, and sulfide uptake rate. The parameters underlying the population growth model were refined using growth

63 data from an additional 615 individuals and population data from an additional 11 aggregations comprised of 4908 individuals. The model presented here includes data from a total of 23 tubeworm aggregations from 3 nearby sites (Green Canyon oil lease blocks 184, 232 and 234) on the upper Louisiana slope to arrive at generalized population growth parameters. Previous studies have shown that L. luymesi has an average longevity of 135 years (Chapter 2), and requires an average of 210 years to reach 2m in length

(Bergquist et al. 2000) a size not uncommon among collected animals. Mortality events are exceedingly rare, dropping below 1% annual mortality probability for animals over

30 cm (Chapter 2). The expanded data sets of growth and mortality rates included here extend the longevity estimate for L. luymesi to an average of 176 years and the estimated age of a 2 m long animal to 216 years.

At the beginning of each iteration, population growth parameters are chosen for the following population growth model:

dN aN = c (1) dt ⎛ aN ⎞ 1+ b⎜ ⎟ ⎝ K ⎠ where N = population size, t = time (years), K = carrying capacity (set to 1000 individuals for all simulations presented here), a describes the initial slope of the line, b defines the degree of density dependence and c is a shape parameter. The first parameter (a) was generated using the following function:

a = 0.7451 + (0.444 · ε[N(0,1)]) (2)

64 where ε[N(0,1)] is a normally distributed random deviate with average = 0 and standard deviation = 1. The other parameters were not normally distributed, and therefore the log- transformed distributions were used to define the distribution of the random numbers generated. As the three parameters in the model were significantly correlated (ln(a) vs. ln(b) r = -0.853, p < 0.001; ln(a) vs. ln (c) r = -0.461, p = 0.036), values of b and c were chosen from their relationship with a:

ln(b) = -6.5309ln(a) + 3.4174 (3)

ln(c) = -0.3789ln(a) + 1.3561 (4)

The value of a was allowed to vary each year according to the pooled standard error associated with the estimates of a from the empirical data (se = 0.105). Once population size equaled or exceeded carrying capacity, recruitment was ceased, representing the lack of additional substrate or sulfide available in the water column.

Once recruitment was determined for that year, the individual-based portion of the model began. Each individual was traced through time with respect to its length, root length, mass, mortality probability, sulfide uptake rate, sulfate excretion rate, and hydrogen ion elimination rate. Individual growth rate was determined from the following non-linear function:

dl = 4.554e −0.01269l ± [2.007e −0.00537l ⋅ ε(N(0,1))] (5) dt

Length (l) is defined here as the distance from the anterior end of the tube to an outer tube diameter of 2mm following the methodology of Bergquist et al. (2000). This

65 function is derived from empirical data on the increase in size of individuals which were stained in situ and collected 12-14 months later. All growth rates were standardized to

365 days. The error term is an additional function fitted to the residuals of the first nonlinear regression function, resulting in a variable growth rate (Fig. 3-2).

10 8 9 6 8 7 4 6 2 5

growth 0 4 abs(residual) 3 -2 2 -4 1 0 -6 0 50 100 150 200 250 0 50 100 150 200 250 length length

Figure 3-2: Growth model for L. luymesi. Different colors indicate growth data from different tubeworm aggregations. Blue points are all from Bergquist et al (2000) a: Average and 95% confidence interval of size specific growth of individuals. b: Error function fitted to the residuals of the model in a.

The ratio of root length to tube length was determined from individual length from the following function:

r : s = 6.134l −0.7024 +1.0 (6)

Mortality probability was determined from the frequency of empty tubes

(representing recent mortality events) within a 10cm size class using the following function:

66 m = 0.0298e −0.0446l (7) where m = mortality probability and l = length. Individuals were considered dead if their probability exceeded a uniform random number between 0 and 1. Individual sulfide uptake was allowed to vary within the range of laboratory determined sulfide uptake rates according to that individual’s growth rate for that year:

⎡ ⎛ g ⎞⎤ u = m⎢1.60 + 4.40⎜ ⎟⎥ (8) ⎣ ⎝10 ⎠⎦ where u = uptake rate (µmol·g·hr), m = mass (g), and g = growth rate (cm·yr-1). Growth rate was divided by the maximum growth rate (10 cm·yr-1) such that highest growth rates resulted in highest uptake rates. The amount of sulfate which could be excreted by each individual was determined from the amount generated by sulfide oxidation carried out by the internal chemoautotrophic symbionts assuming constant internal sulfate concentration. Hydrogen ions are also generated in the oxidation of sulfide by the tubeworm symbionts. Hydrogen ion elimination rate was determined in the same fashion as sulfide uptake, with growth rate determining the variability in this metabolic flux according to laboratory measured ion fluxes (mean = 10.96 µmol·g-1·hr-1, sd = 1.88

µmol·g-1·hr-1) (Girguis et al. 2002). Simple diffusion of hydrogen ions across the root surface was included in the model. As this accounts for a relatively small proportion of total ion flux (less than 10% in large individuals), additional pathways are likely and require further investigation.

67 Geochemical setting

Known sources of sulfide available to L. luymesi aggregations are sulfide transported with seeping fluids (Carney 1994) and sulfide generated via reduction of seawater sulfate (Aharon and Fu 2000, Aharon and Fu 2003). Putative sources of sulfide associated with seepage include anaerobic oxidation of deeply buried organic material

(Carney 1994), of “sour” hydrocarbons containing a proportion of sulfur (Kennicutt et al.

1992), and hydrocarbon interactions with sulfur-bearing minerals such as gypsum and anhydrite found in the salt dome cap rocks of the ULS (Sassen et al. 1994, Saunders and

Thomas 1996).

Concentrations of all chemical species in the sediments surrounding the rhizopshere were derived from the data sets included in Morse et al. (2002) and Arvidson et al. (2004). Only those sediment cores taken around the “drip line” of tubeworm aggregations which contained detectable sulfide concentrations were used. Due to the vagaries of sampling with a submersible in sediments heavily impacted by carbonate and roots, those cores with detectable sulfide are believed to more accurately represent to conditions around the periphery of the rhizosphere.

Dissolved organic carbon (DOC) concentration was used as an estimate of methane concentration. While other forms of dissolved organic carbon make up this total concentration, methane accounts for 90-95% of the hydrocarbon gasses dissolved in pore waters (Morse et al. 2002). In seep sediments, the majority of dissolved organic carbon is likely to be in the form of hydrocarbon gasses. Because estimates of organic acid concentrations were not available, they could not be explicitly modeled. This would not

68 affect the overall concentration of electron donors in the model, but could affect the sulfate reduction rate. Since sulfate reduction rate estimates for methane seeps in the Gulf of Mexico are among the highest recorded, any differences in DOC composition would serve to lower the overall sulfate reduction rate and sulfide availability. Therefore, the sulfide supply estimates presented are likely overestimated, particularly for the model without root sulfate release due to the greater reliance on anaerobic methane oxidation in this simulation.

Solid and liquid phase organic carbon was separated into hydrocarbons and buried organic material according to their relative concentrations in hydrocarbon seep and surrounding Gulf of Mexico sediments. Background sediments on the upper Louisiana slope (ULS) contain 0.71% organic carbon by weight (Lin and Morse 1991). At hydrocarbon seeps on the ULS, organic carbon accounts for 4.47% of total weight. This was assumed to be the sum of background organic input plus carbon in the form of C6+ hydrocarbons. It is possible that the higher biomass located at ULS seeps in the form of non-living macrofaunal and microbial materials may also contribute to the increased organic carbon concentration, but without empirical estimates, this could not be accounted for in the model. Hydrocarbons may consist of between 50 and 95% labile materials (Mana Capelli et al. 2001, Delille et al. 2002, Mills et al. 2003). Based on existing data on degradation rates and residual hydrocarbons subjected to degradation

(Kennicutt et al. 1988, Sassen et al. 1994), a value of 50% labile material was used here.

These assumptions of hydrocarbon concentration and degradation potential are therefore believed to be conservative.

69 The following nonlinear functions were fitted to the sulfide, sulfate and methane concentration profiles to determine the boundary conditions at any given depth:

−ad Ci = ()Co − C∞ ⋅ e + C∞ (9)

where C0 is initial concentration, C∞ is concentration at infinite distance, Ci is concentration at depth = d ( Fig. 3-3 ). As there was no existing data for sediments below

2- 30cm, concentrations at infinite depth (C∞) were used (SO4 = 0, HS- = 12mmol, DOC =

11mmol, DIC = 20mmol, pH = 7.78).

sulfide (mmol) sulfate (mmol) DOC (mmol) 024601020300 5 10 15 20 0

-5 )

-10

-15 sediment depth (cm -20

-25

Fig. 3-3: Concentration profiles of sulfate, sulfide, and dissolved organic carbon (DOC). Points represent average concentration at a given depth from 13 sediment cores taken around the periphery of tubeworm aggregations. Best fit line based on least squares fit of equation 9.

The first derivatives of the sulfide and methane profiles were used for the calculation of advective flux. Advection (seepage) rate varied with time according to the following function:

70 dz = 0.3649e−0.157t + 0.000365 − sed (10) dt where t = simulation time in years and sed is sedimentation rate (6cm·1000yr-1) (Lin and

Morse 1991). Early seepage rate approximated the highest flux rates measured or estimated for methane seeps and declined with time in the model to the highest estimates for persistent, region-wide seepage in the Gulf of Mexico ( Table 3-1 ). This follows a pattern of hydrocarbon seep development with the highest seepage rates early in the evolution of the local seepage source followed by occlusion of seepage pathways by carbonate precipitation, hydrate formation, and possibly tubeworm root growth. By using the highest rate estimated (35mm·yr-1) as the basal seepage rate, we are testing the possibility that tubeworm aggregations could survive under the most favorable conditions possible.

71 Table 3-1: Reported seepage rates for hydrocarbon and methane seeps. 1 MacDonald et al. 1993, 2 MacDonald et al. 2000, 3 Rudinicki et al. 2001, 4 Carson et al. 1991, 5 Henry et al. 2002, 6 Olu et al. 1997, 7 Wallmann et al. 1997, 8 Levin et al. 2003, 9 Wiedicke et al. 2002, 10 Peacock 1991. site published mm/yr notes

Gulf of Mexico 1.4 m3/1000km2/day 1.5 region wide seepage1

Gulf of Mexico 30 m3/1000km2/day 32 region wide seepage1

Gulf of Mexico 7km/1Myr 7 oil migration2

Juan de Fuca 2 mm/yr 2 "strong upwelling"3

Juan de Fuca 4000m/Myr 4 maximum of modeled rates3

Cascadia 5*10-12 m3/m2/sec 0.26 vertical compaction model4

Cascadia 10-6 m3/m2/sec 52000 measured at "small" vent4

Nankai 10 to 30 m/yr 20000 temperature profiles at active vent5

Barbados 0.2 to 10 mm/yr 5.1 temperature profiles (Darcy Velocity)6

Barbados 6 cm/yr 60 temperature profiles (Darcy Velocity)6

Aleutians 3.4 m/yr 3400 to support benthic oxygen flux7

Eel River, CA 10 cm/yr 100 in situ measurement8

Indonesia 0.15 m/yr 1500 expulsion of warm fluids9

Indonesia 3.1*10-5 m/yr 0.31 "marginal area" near vent9

N.A. 1 kg/m2/yr 1.61 theoretical10

For sediments encompassed by the rhizophere, sulfide, sulfate, methane, DOC and hydrogen ion concentration profiles were determined iteratively prior to model implementation using a central difference scheme:

72

⎛ C ⎞ ⎜ i ⎟ Ci(t+1) = Ci + D()Ci+1 − 2Ci + Ci−1 − k⎜ ⎟ (11) ⎝ K s + Ci ⎠

where Ci(t) = concentration in cell i at time t, D = diffusion coefficient, k = maximum reaction rate, Ks = half saturation constant for the reaction. Reactions included anaerobic methane oxidation (including sulfate reduction, bicarbonate generation and hydrogen ion consumption), tubeworm sulfide uptake rate, and carbonate precipitation rate. The model calculated the concentration in each 2 cm by 2 cm cell at 1 hour time steps. At the end of each year, diffusion distance increased. The number of cells (total diffusion distance) was determined by the average root length of L. luymesi populations as realized in independent runs of the population growth model described above, and included here as model input only. A separate non-linear function was fit to each of the concentration profiles:

−ad Ci = ()C0 − C∞ e + C∞ (12) where d is radial distance. The relationship between the parameter a and distance was used to generate concentration profiles for each disc comprising the rhizosphere. Because of the tight linear relationship between diffusion distance and the shape of the curve, concentration profiles could be generated for a disc of any size using the following function:

−β ahs = αd (13)

73 - 2- where α = 1.7344 and β = -1.0104 for HS , α = 0.2111 and β = -0.3363 for SO4 , and α =

0.1626 and β = -0.2518 for CH4. Diffusional fluxes of sulfide, sulfate, and methane were calculated according to the first and second derivatives of the concentration profiles as determined by the diameter of each disc.

Model implementation

The region encompassed by the model (the rhizosphere) was an inverted dome with a diameter equal to the average root length of the population. To reduce the complexity of the solution of diffusional flux across a 3-dimensional spherical object, this shape was approximated by a series of 2cm discs (Fig. 3-1). Diffusional fluxes into each disc were calculated from the shape of the concentration profiles according to the following function (Boudreau 1997): dC 1 d ⎛ dC ⎞ = ⋅ ⎜rDs ⋅ ⎟ (14) dt r dr ⎝ dr ⎠ where C = concentration, r = disc radius, and Ds = diffusion coefficients corrected for porosity by:

D D = o (15) s 1+ n(1− φ) where Do = diffusion coefficient corrected for temperature and pressure, n = chemical species-specific constant, and φ = porosity. The value of n was set to 2.75 as this was found to be a reasonable fit for all chemical species examined (Iversen and Jorgensen

1993). The ionic states of each species at the average pH value of tubeworm-dominated

74 sediments (7.78) were used for the determination of diffusion coefficients. Porosity was determined from the following function:

−az φ z = ()φ0 − φ∞ e + φ∞ (16)

φz = porosity at depth z, φ0 = porosity at sediment/water interface, φ∞ = porosity at

infinite depth, φ0 = 0.841, φ∞ = 0.765 and a = 0.210 as determined from the best fit with the porosity data ( Fig. 3-4 ) from Morse et al. (2002).

poros ity 0.75 0.8 0.85 0

-5 )

-10

-15

-20 sediment depth (cm -25

-30

Fig. 3-4: Sediment porosity values. Points represent average porosity at a given depth from 13 sediment cores taken around the periphery of tubeworm aggregations. Best fit line based on least squares fit of equation 9.

Diffusion across the sediment/water interface of the rhizosphere was also considered as an additional input of sulfate and hydrogen ions. This was included as one- dimensional diffusion across a circular surface (subtracting the area encompassed by the tubeworm tubes) with diffusion distance equal to rhizosphere diameter, and concentration differential from seawater concentration to average concentration within the rhizosphere.

Sulfate and hydrogen ion diffusion across the root surface was then added (if included in

75 the set of model realizations) as simple Fickian diffusion. Concentration differential was the difference between internal concentration and average concentration for each disc of the rhizosphere assuming roots were evenly proportioned according to the volume encompassed by each disc. Internal sulfate concentration and pH (Table 3-2) represented an average of the values determined Riftia pacyptila (Goffredi et al. 1999), a hydrothermal vent tubeworm. Internal sulfate concentrations and pH of L. luymesi have not been reported, but are generally consistent within taxa (Schmidt-Nielsen 1997).

76 Table 3-2: Parameters involved in diagenetic model. Diffusion coefficients and disassociation constants corrected for temperature (t), salinity (s) and pressure (p) where noted according to Pilson (1998) and Stumm and Morgan (2000). 3-2a: Diffusion coefficients (cm2·sec-1·10-5). All corrected for temperature and pressure. 3-2b: Disassociation constants. All corrected for temperature, salinity and pressure except CaOH (none), CaHCO3 (t), CaSO4 (t), CaSO4H2O (t), MgHCO3 (t), H2CO3 (t,s), and 3 -1 HSO4 (t,p). 3-2c: Maximum reaction rates (k) (µM·cm ·sec ) for sulfate with methane (CH4), larger hydrocarbons (HC), and organic matter (G). 3-2d: Half-saturation constants (Ks) (µM).

a c Do k - -5 HS 1.230 SO4 + CH4 2.65·10

2- -6 SO4 0.650 SO4 + HC 2.50·10 -8 CH4 1.021 SO4 + G 4.90·10

- HCO3 0.726 b d H+ 6.684 Ks

- HS 85.8 K* 2- SO4 1500 -7 H2S 1.310x10

- HSO4 6.354

-7 H2CO3 8.154x10

-10 HCO3 4.727x10

-9 B(OH)3 1.634x10

H2O 1.320

CaOH 0.040

CaHCO3 8.722

-7 CaCO3 5.043x10

-5 CaSO4 4.584x10

-5 CaSO4H2O 2.538x10

MgHCO3 11.203

77 Within the rhizosphere, sulfide generation may be limited by sulfate supply, electron donor availability, or sulfate reduction rate. Sulfate supply was determined as the sum of flux across the series of discs approximating the rhizosphere dome, across the sediment-water interface, and from root sulfate (if available). Electron donors included methane, larger hydrocarbons, and buried organic material. Sulfate reduction rate was determined from the relative amounts of the various electron donors with higher rates

(0.7132 µmol·ml-1·h-1) for methane oxidation and lower rates (0.0832 µmol·ml-1·h-1) for organic matter or hydrocarbon degradation (Aharon and Fu 2000).

Total hydrogen sulfide availability to the aggregation was determined as the sum of sulfide diffusion and advection into the rhizophere and sulfide generated within the rhizosphere from sulfate reduction according to the following reactions:

2- - - SO4 + CH4 → HS + HCO3 + H2O (17)

2- - - SO4 + 2CH2O → HS + 2HCO3 + H2O (18)

2- - - SO4 + 1.47CnH2n+2 → HS + 1.47HCO3 + H2O (19)

- bicarbonate (HCO3 ) is generated at a 1:1 stoiciometry during anaerobic methane oxidation and a 2:1 stoiciometry in the degradation of organic material. As hydrocarbons are degraded forming smaller chain hydrocarbons and organic acids, bicarbonate is generated at different stoiciometries. Because different sized hydrocarbons and organic acids were not accounted for in the model, a rough average of these stoiciometries

(1.47:1) based on toluene, ethylbenzene, xylene, and hexadecane degradation (Zwolinski et al. 2000) was used to determine the amount of bicarbonate generated per mole of carbon. Hydrogen ions are also used up in a 1:1 stoiciometry with sulfate in the sulfate reduction half reaction as included in reaction 17.

78 In order to account for carbonate precipitation, the model traced dissolved inorganic carbon (DIC) concentration, calcium concentration, hydrogen ion concentration, buffer capacity, carbonate saturation, and carbonate precipitation rate. The buffer state of the rhizosphere was calculated to determine changes in pH resulting from hydrogen ion flux. Buffer capacity was calculated using the following function (Stumm and Morgan 2000):

⎡ + − ⎛ []A [B] ⎞ ⎛ [A][B] ⎞ ⎤ B = 2.3⋅ ⎢[][H + OH ]+ ⎜ ⎟ + ... + ⎜ ⎟ ⎥ (20) ⎣ ⎝ [][]A + B ⎠1 ⎝ [][]A + B ⎠ n ⎦ where A and B represent the concentrations of the various acids and bases in the buffer system. In addition to hydrogen and hydroxy ions, the buffer system included carbonate

- 2- - - 2- (CO2, H2CO3, HCO3 , and CO3 ), sulfide (H2S and HS ), sulfate (HSO4 , SO4 ), and

- borate (B(OH)4 , B(OH)3) speciation. Current pH was used to determine the ionic state of each species according to temperature, pressure and salinity corrected disassociation constants when available (Pilson 1998, Stumm and Morgan 2000) (Table 2). Change in pH was determined from hydrogen ion flux and buffer capacity as follows:

d[H + ] dpH − = dt (21) dt B

Saturation state is highly dependent on the degree to which calcium and bicarbonate form complexes with other ions. The “free” calcium was determined as the

- proportion of calcium which is not associated with complexed bicarbonate (HCO3 ),

2- - 2- carbonate (CO3 ), hydroxyl (OH ), or sulfate (SO4 ) ions. Free carbonate was determined as the amount not forming complexes with calcium (Ca+) or magnesium (Mg+) ions in solution. Saturation state was then calculated form the product of the concentrations of

79 free calcium and carbonate divided by the solubility product constant. If the saturation state was above 1, then carbonate precipitation occurred at a rate determined by:

d[]CaCO3 + − + 2− = k1 [Ca ][]HCO3 + k3 [Ca ][CO3 ] (22) dt

4 -1 -1 4 -1 -1 where k1 = 0.00597 cm ·mmol ·sec and k3 = 0.456 cm ·mmol ·sec (Stumm and

Morgan 2000). Because of a lack of an empirical relationship between weight percent of carbonate and sediment porosity in tubeworm dominated sediments (Morse et al. 2002), precipitation did not directly affect porosity. Precipitation was accounted for in the model by subtracting the volume of carbonate precipitate from the total volume encompassed by the rhizosphere.

At the end of each annual time step, model output included average length of individuals, population size, sulfide uptake rate, sulfide supply rate, root sulfate flux (if included), root hydrogen ion flux, amount of sulfide supply accounted for by each process (sulfide seepage, anaerobic methane oxidation, organic matter degradation, and hydrocarbon degradation), number of individuals which could be supported by sulfide supply, carbonate precipitation rate, volume of carbonate precipitate, and pH.

Results and Discussion

The model predicts that inputs from known sources, including diffusion and advection of deep sulfide along with reduced seawater sulfate (Fig. 3-1), will support a moderately sized aggregation of 1000 individuals for an average of 39 years (range: 22 to

78 years) (Fig. 3-5). The maximum amount of time that any one realized aggregation

80 persisted in the model (78 years) represents an aggregation with the most rapid growth and slowest recruitment possible, as determined by the distribution of empirical data on tubeworm population growth. This suggests that only aggregations with population growth characteristics outside the distribution of those so far observed in the 23 aggregations collected over the past 7 years would persist beyond 80 years, though this possibility cannot be entirely dismissed. A smaller aggregation of 200 individuals could be maintained with these sources for an average of 64.1 years (sd = 10.6 yrs). In this model configuration, the duration of adequate sulfide flux is not congruent with existing age estimates of L. luymesi individuals and aggregations.

100000

10000

d 1000

100

supply:deman 10

1

0.1 0 50 100 150 200 time (years)

Fig. 3-5: Ratio of sulfide supply to sulfide uptake rate of L. luymesi aggregations based on known sources without sulfate release by tubeworm roots. Sulfide supply declines below demand after approximately 40 years. Equilibrium line (1:1 ratio) and average, maximum and minimum values for 1000 iterations presented.

The addition of sulfate release by tubeworm roots results in modeled sulfide generation and flux at rates which match the demands of large aggregations, allowing the

81 tubeworms to survive for over 200 years (Fig. 3-6). This additional source of sulfate results in a 2 orders of magnitude increase in sulfate flux in older (>100 yrs) aggregations, accounting for over 90% of sulfate available after only 24 years. It should be noted that in the model, sulfate release is constrained by the rate of sulfate generation by the tubeworm’s sulfide oxidizing symbionts (resulting in the near 1:1 ratio of supply and demand in Fig. 3-5). The majority of sulfate supplied by tubeworm roots is utilized for microbial hydrocarbon degradation ( Fig. 3-7 ). This process alone accounts for over

90% of the sulfide available to aggregations after approximately 80 years. By augmenting the sulfate supply to microbial consortia for anaerobic methane oxidation and hydrocarbon degradation, large aggregations of tubeworms may survive for hundreds of years, resulting in the population sizes and individual lengths regularly observed and collected at seeps in the Gulf of Mexico (Bergquist et al. 2003).

82

100000

10000 d 1000

100

supply:deman 10

1

0.1 0 50 100 150 200 time (years)

Fig. 3-6: Ratio of sulfide supply to sulfide uptake rate of L. luymesi aggregations including sulfate release from tubeworm roots, with sulfate release constrained by tubeworm symbionts sulfide oxidation rate. Sulfide supply exceeds demand for the duration of the model. Equilibrium line (1:1 ratio) and average, maximum and minimum values for 1000 iterations presented.

83

1.0 0.9 0.8 y 0.7 hydrocarbons 0.6 organic matter 0.5 methane 0.4 sulfide

proportion of suppl of proportion 0.3 0.2 0.1 0.0 0 20 40 60 80 100 120 140 160 180 200 time (years)

Fig. 3-7: Sources of sulfide available to tubeworm aggregations. Sources of sulfide include advection and diffusion of sulfide from deep sources (yellow) or sulfate reduction using methane (blue), buried organic carbon (green), or C6+ hydrocarbons (grey) as electron donors. Sulfate is provided by diffusion from sediments surrounding the rhizosphere, diffusion at the sediment/water interface, and sulfate released from tubeworm roots. Results indicate an initial reliance on advective sulfide flux, with heavy reliance on sulfide produced by anaerobic hydrocarbon degradation in older aggregations.

An alternate hypothesis to explain the discordance between estimated supply and uptake rates is the presence of locally elevated seepage rate. Sensitivity analysis was carried out to determine the potential effects of uncertainty in seepage rate on supply estimated for aggregations without root sulfate release. A 10% increase in seepage rate resulted in a 5.6% increase in sulfide supply to 200+ year old aggregations. This corresponds to only 16.4% of the sulfide required, which does not serve to extend aggregation survivorship (average = 39 yrs, range = 21 to 79 yrs) beyond that determined for lower flow rates. To supply the sulfide flux required by older aggregations, seepage rate would have to be at least 363 mm·yr-1. This is over 10 times greater than the rate

84 used in the model (32 mm·yr-1), which is the highest region-wide estimate for the Gulf of

Mexico (MacDonald et al. 1993) and approaches rates reported for active venting of fluids (Table 1).

The release of sulfate by tubeworm roots potentially explains the frequent observation of highly degraded hydrocarbons in the vicinity of large tubeworm aggregations (Sassen et al. 1999). By increasing sulfate flux to deeper sediments, L. luymesi increases integrated rates of anaerobic hydrocarbon degradation, leading to highly biologically altered hydrocarbon pools among the roots of tubeworm aggregations.

Hydrocarbon oxidation has been implicated as one of the dominant processes in the carbon cycle at ULS seeps, accounting for over 90% of the carbon in carbonates collected in the vicinity of tubeworm aggregations (Formolo et al. 2004). Model analysis indicates that the minimum amount of organic carbon in sediments required to maintain a 1:1 supply:uptake ratio is 1.03% by weight, remarkably close to the lowest value found in any of the seep sediment core samples (1.2%) (Morse et al. 2002, Arvidson et al. 2004), and greater than that found in ULS sediments away from seeps (0.71%) (Lin and Morse

1991). Elevated organic carbon content at seeps, primarily resulting from oil seepage, provides the energy source required to generate sufficient sulfide for tubeworm aggregations. This previously overlooked source of sulfate would explain the high apparent diffusion coefficients determined for tubeworm-impacted sediments (Arvidson et al. 2004). The hypothesized release of sulfate by tubeworm roots potentially explains numerous, apparently disparate observations, hinting at the great impact that L. luymesi aggregations may have on their abiotic environment.

85 Tubeworm sulfate release would further alter the biogeochemistry of seeps by enhancing calcium carbonate precipitation, a by-product of increased integrated rates of hydrocarbon degradation and methane oxidation. In order to prevent the precipitation of carbonate on the root surface, which has not been observed on L. luymesi (pers. obs.), L. luymesi individuals may release hydrogen ions as well as sulfate through their roots. In the model, diffusion of hydrogen ions across the root surface (the only form of release explicitly modeled) accounts for less than 40% of ion generation when carbonate precipitation is most vigorous. L. luymesi may utilize the excess hydrogen ions generated by their sulfide oxidizing symbionts to periodically raise the rate of H+ flux from their roots. This would not only supply additional hydrogen ions to sulfate reducing bacteria, but could inhibit local carbonate deposition and subsequent reduction of the root area utilizable as a respiratory surface. Further pH reduction could dissolve existing authigenic carbonate thereby opening seepage pathways and allowing further root growth. This possibility is corroborated by the observation of young tubeworms which had apparently bored through bivalve shells in an experimental system (pers. comm. R. Carney, L.S.U.).

Empirical measurements of hydrogen ion flux across the root tissue of L .luymesi are required to test this hypothetical mechanism.

While model results indicate that the proposed relationship between symbiotic tubeworms and sulfate-reducing microbial consortia is essential for the longevity of L. luymesi and persistence of aggregations, there are significant effects on the microbial community as well. Integrated anaerobic methane oxidation and hydrocarbon degradation rates will increase, resulting in greater microbial biomass. Tubeworm generated sulfate supplies a more energetically favorable electron acceptor, relaxing the limitation on

86 anaerobic oxidative processes below the normal depth of sulfate penetration at seeps.

Deeper sediment layers then become habitable to sulfate reducing consortia, significantly altering the microbial community structure at seeps. Model configurations neglect the potential role of bio-irrigation of seawater sulfate through L. luymesi tubes which could further increase sulfate supply to deeper sediment layers. The possible role of tubeworm roots as substrata for the growth of microbial consortia, analogous to the habitat afforded mycorrhizal symbionts of higher plants, remains another possible benefit for the microbes.

Model results support the hypothesis that a complex relationship between an animal with bacterial endosymbionts and external bacterial/archaeal consortia results in the persistence of the longest-lived animal known. This positive interspecific relationship benefits both the tubeworms and the microbial consortia involved. This finding expands our existing concept of the potential for complexity in symbioses and the benefits they may confer. Further complex relationships are likely to be discovered through continued research into the role of positive species interactions at the individual and community level.

Notes

Michael Arthur, Kat Shea, Rolf Arvidson and Chuck Fisher are coauthors on this manuscript, which is under review at PLoS: Biology. They all assisted with writing and revising early versions of the manuscript. Michael Arthur assisted with the construction of the model. Katriona Shea assisted with the programming and implementation of the

87 model. Rolf Arvidson provided the geochemical data that formed the basis of the model.

Chuck Fisher originally posed the question of supply rates to tubeworm aggregations and assisted in refining the model. Derk Bergquist provided data on tubeworm aggregation characteristics from previous studies. Ben Predmore, Meredith Redding, Julie Barsic,

Brian Tiegs, and Nicole Iacchei all contributed to the enumeration and measurement of tubeworms.

Chapter 4

Succession of Hydrocarbon Seep Communities Associated with the Long-Lived Foundation Species Lamellibrachia luymesi

Abstract

The vestimentiferan tubeworm, Lamellibrachia luymesi, lives for over 250 years and forms aggregations of hundreds to thousands of individuals at hydrocarbon seeps in the Gulf of Mexico. A model of community succession within this biogenic habitat has been proposed where changes in the abiotic environment lead to shifts in the structure of tubeworm-associated communities. Here, we validate this model by examining succession in the communities associated with 11 tubeworm aggregations. Young aggregations contain high biomass communities comprised primarily of endemic species in lower trophic levels. As aggregations age and sulfide levels decline, rate of endemism, overall biomass, and density of fauna decline, while relative dominance of higher trophic levels increases. Diversity of the community appears to increase initially and then decline in much older (>100 years) aggregations. Successional patterns result from temporal patterns of sulfide abundance affecting local primary productivity and excluding non- endemic species. The dominance of higher trophic levels in older aggregations is enabled by the connectivity between adjacent habitats.

89 Introduction

The impact of positive species interactions has received little attention in the development of ecological theory (Bruno et al. 2003). These interspecifc interactions allow the species involved to colonize and persist in previously exclusionary habitats, expanding their ecological niches (Higashi 1993). While the effects of these interactions on the species directly involved have been demonstrated, the ecological impact of these relationships on the surrounding community is less well understood (Stachowicz 2001).

Though often overlooked, the creation of physical structure by foundation species can be the most significant force affecting community structure on the landscape scale (Hacker and Gaines 1997). The positive species interactions implicit in the creation of biogenic habitat can lead to elevated diversity as compared to the surrounding, unaltered habitat

(Hacker and Gaines 1997). Forests and coral reefs harbor some of the most diverse communities known in terrestrial and marine systems (Connell 1978). This type of positive interaction has repercussions well beyond the species directly involved, impacting entire communities of organisms (Bertness and Leonard 1997).

The modification of the abiotic environment can further impact the community structure of biogenic habitats. The alteration of existing habitat will inhibit the continued persistence of the present community, while facilitating colonization by other species

(Harris et al 1984). This transitional period can be followed by a stable state persisting on long temporal scales. Ponds created by beaver exist for decades (Naiman et al 1986).

Forests may persist for centuries, while coral reef structures endure for millennia. If the

90 effects of habitat modification are manifest in a predictable and consistent sequence, these are considered successional processes (Odum 1969).

The communities associated with biogenic habitats at hydrocarbon seeps on the continental slope of the Gulf of Mexico have been proposed to follow such a sequence of successional changes (Bergquist et al. 2003). These communities are dominated in terms of biomass by vestimentiferan tubeworms and mytilid mussels with internal chemoautotrophic or methanotrophic symbionts, respectively (Brooks et al. 1987,

MacDonald et al. 1990). Associated with these foundation species are faunal assemblages which are far more diverse than those found in surrounding benthic habitats (Carney

1994). A model of succession was proposed by Bergquist et al. (2003) which involved a sequence of 4 community stages replaced over the course of centuries. These communities begin with mussel (Bathymodiolus childressi) beds containing high biomass communities of low diversity and high endemicity. In this first stage, the relatively high levels of methane and sulfide result in high microbial productivity, but also produce an exclusionary environment favoring endemic species with high sulfide and hypoxic tolerances (Bergquist et al. in press).

The next three successional stages are dominated by vestimentiferan tubeworms.

The mussel bed stage overlaps with and is succeeded by the second stage, “juvenile” tubeworm aggregations composed of two species of vestimentiferans, Lamellibrachia luymesi and Seepiophila jonesi. The abiotic environments of the mussel beds and young tubeworm aggregations contain high levels of reduced chemicals, giving rise to similar communities with different foundation species creating the habitat. The third stage involves a reduction of sulfide and methane concentrations in the epibenthos leading to

91 increased colonization by non-endemic species with presumably lower tolerances to sulfide and hydrocarbons. This results in lower total biomass of associated fauna due to a decline in productivity, but higher overall diversity with a mixture of endemic and non- endemic species.

As epibenthic sulfide levels decline further, microbial productivity declines to very low levels leading to further reductions in biomass of tubeworm-associated fauna.

Low productivity combined with increased predation act in conjunction to nearly eliminate lower trophic levels from the community. Colonization of non-endemic sedentary filter feeders increases, at least visually, resulting from decreased sulfide as well as lower grazer abundance. At this point, the large physical structure created by the tubeworms largely becomes a complex extension of the surrounding habitat. This final stage may last for decades to centuries, as the tubeworms often live for over 250 years.

This model of succession integrated all that was known of the natural history of the vestimentiferans and the community structure of the associated fauna. The majority of the information used in this previous study was gleaned from 4 quantitative and 3 qualitative collections from 2 sites, using the visual appearance of aggregations to place them in a temporal sequence. Here, we test this model in a mensurative experiment using

11 quantitative samples of tubeworm communities, including 3 collections from a site which has not been previously sampled in this manner. In addition, we use a population growth model to estimate the ages of each aggregation in order to quantify the temporal scale of the hypothesized successional processes.

92 Methods

A total of 13 vestimentiferan aggregations were collected on 3 cruises aboard the

R/V Seward Johnson II in 2002, 2003, and 2004. Collections were made at 2 sites on the upper Louisiana slope (ULS) of the Gulf of Mexico (Fig. 4-1). The sites were approximately 10 km apart in the Green Canyon (GC) 234 (site 1) and GC232 (site 2) lease blocks of the ULS and ranged from 528 to 571 m depth. These sites appear to have similar geochemical settings, including similar hydrocarbon gas composition (Sassen et al. 1994) and hydrogen sulfide concentrations (Aharon and Fu 2000, Aharon and Fu

2003), as well as the presence of methane hydrate breaching the seafloor (Sager et al.

2003, pers. obs.).

Figure 4-1: Map of collection sites in the northern Gulf of Mexico. Depth contours are in meters

93 For determination of the tubeworm-associated community structure, vestimentiferan aggregations were obtained using the Johnson Sea-Link I and II submersibles with the Bushmaster Jr. and Bushmaster Sr. collection devices. The

Bushmaster collection devices are hydraulically actuated nets lined with a 63µm mesh.

Tubeworm aggregations were collected by placing one of the devices over the aggregation and constricting a stainless-steel cable around the base. The aggregation was removed from the sediment and the intact collection placed in a receptacle mounted to the front of the submersible. For Bushmaster Jr. collections, the receptacle was the bottom half of a 30 gallon drum lined with 64 µm mesh. In Bushmaster Sr. collections, the device was placed on a 1.8 m diameter trampoline mounted to the front of the submersible. Once the submersible was retrieved, the Bushmaster was removed from the receptacle and the aggregation placed into a plastic tub on the deck of the ship. All contents of the Bushmaster and the receptacle were washed into the tub. At this point, the associated fauna was separated and the tubeworms removed from the tub. The contents of the tub were sieved through a 2 mm mesh and sorted on board to lowest possible taxonomic level. This mesh size was chosen due to the large volume of sediment collected with the aggregations and the limited time available during the cruise for processing. Subsamples of smaller size-fractions were retained for future small macrofaunal and meiofaunal community investigations.

All associated fauna were preserved and transported back to Penn State for final determination of taxonomic status. E.E.C. and S.H. identified all polychaetes. Primary identification of other groups was carried out and specimens sent to experts for further identification or verification. All individuals were weighed and enumerated for

94 subsequent analyses. Preserved wet weight was converted to ash free dry weight

(AFDW) using existing species specific conversion factors (Bergquist et al. 2003a) when available, or by published higher taxonomic level conversion factors (Ricciardi and

Bourget 1998). Species were assigned a trophic level based on published species-specific trophic relationships when possible (see Bergquist et al. 2003a for list of references), or reported family- or higher level trophic interactions. Species are reported as endemic if they have only been collected at cold seeps or contain chemosynthetic or methanotrophic symbionts. Community diversity of each collection was assessed using the Shannon-

Weaver diversity index (H’):

H '= − p log( p ) (1) ∑i i i where pi is the relative abundance (%) of the ith species. Community evenness was determined using Pielou’s index of evenness (J):

H ' J '= (2) log S where S = total number of species in all collections. Colonial organisms were not enumerated, and therefore were not included in quantitative analyses.

Tubeworms were counted and measured on board, time permitting. The remainder of the aggregations were preserved in formalin and transported back to Penn State for processing. Length was measured to a standardized posterior outer tube diameter, 2mm for L. luymesi and 4mm for S. jonesi. This diameter was the common point where the tubeworms entered a dense, tangled mass at the base of the aggregations. Efforts to measure the species to a smaller diameter resulted in a significant loss of quantitative data on tube length due to tube breakage. Surface area was calculated as for a cone frustrum:

95 1 SA = π ⋅l()AD 2 + PD 2 + AD ⋅ PD (3) 3 where SA = surface area, l = length, AD = anterior diameter, and PD = posterior diameter.

Published conversions of length to AFDW were used to calculate biomass of tubeworm species (Bergquist et al. 2003a).

Aggregations were assigned an age based on results of a population growth model

(Chapter 2). The model includes population-size dependent recruitment, individual size- specific growth, and individual size-specific mortality for L. luymesi. S. jonesi could not be included in the model due to a paucity of data on their growth and mortality rates. The model was run with aggregation-specific recruitment parameters derived from cohort analysis of size-frequency distributions. If the non-linear recruitment function could not converge on a set of parameter estimates for a given aggregation, then pooled estimates of recruitment parameters were used. The model was run until the average length of tubeworms in the aggregation was within 1 cm of the actual average length. At the end of each model run, the frequency of tubeworms in 10 cm size classes was compared to the actual size-frequency distribution of the collected aggregation using a log-likelihood test.

The 100 best fitting simulations of the 1000 iterations of the model were selected for each aggregation and average age determined from this subset of model realizations.

Hydrogen sulfide was measured using a modification of an enzymatic assay

(Freytag et al. 2001). Water samples were obtained through a 0.5 mm diameter PEEK tubing using the submersible. Within the submersible, samples were collected in 2 ml gas-tight syringes containing 150 uL of a 3.4 mM enzyme (papain) and 20 mM sodium acetate solution. The papain enzyme is activated in the presence of sulfide and forms a

96 complex to prevent further oxidation of sulfide within the syringe. The syringe is kept on ice until retrieval of the submersible when the papain solution is added to a 3.55 mM L-

Bapna solution in EDTA buffer. The activated form of papain cleaves the L-Bapna molecule releasing a p-nitroaniline chromophore with can be quantified using a colorimetric assay on a spectrophotometer. Standard curves were run for the 0.2 to 2.0

µM sulfide concentration range. All sample values under 0.2 µM were not significantly different from 0 (confidence interval of regression overlapped 0), and were therefore treated as 0 in the analysis. Exact values for concentrations determined above 2 µM could not be given with confidence and were therefore treated as a value of 2 µM in the analysis.

Similarity in community structure between aggregations was examined using multidimensional scaling (MDS) using PRIMER software (PRIMER-E ltd.). This analysis uses an iterative procedure to minimize the difference between ranked Bray-

Curtis (BC) similarity values and 2-dimensional distance between each pair of samples.

Prior to determination of pair-wise Bray-Curtis similarity, species abundance was standardized to tubeworm surface area to provide a measure of species density irrespective of collection size. Density data were 4th root transformed to reduce the bias towards the most dominant species (Clarke and Warwick 2001). Bray-Curtis similarity was determined using the following function:

⎛ p ⎞ ⎜ ∑ yij − yik ⎟ Sjk = 100⎜1− i=1 ⎟ (4) ⎜ p ⎟ ⎜ ∑()yij + yik ⎟ ⎝ i=1 ⎠

97 where yij is the abundance of the ith species in the jth sample and p is the total number of species.

To assess the relative influence of abiotic factors on community similarity, a modified BIO-ENV procedure (Clarke and Warwick 2001) was used. Similarity

(Euclidean distance) between collections was determined for each environmental factor.

Spearman rank correlation between Bray-Curtis similarity in community structure and

Euclidean distance in habitat characteristics was used to determine the environmental factor which best explained the pattern in community similarity. Variables tested included aggregation age, sulfide abundance, distance (pairwise Euclidean distance between latitude and longitude), and depth. Age was log-transformed (log(x+1)) to account for the skewed distribution of ages sampled. Sulfide abundance was also log- transformed and entered into the model as concentration at plume, mid, and sediment- level as well as average concentration and proportion of samples with detectable (>0.2

µM) concentrations.

Similarity between species distributions was assessed using a cluster analysis of

Bray-Curtis similarity values for the 25 most common species. Commonality was determined by an index combining each species’ rank in average density, average biomass, and frequency of occurrence in the collections. The 25 species selected in this manner included at least 7 of the 10 most abundant species in each collection and the 10 most abundant species in 5 of the collections. Cluster analysis was based on group average linkage of species densities with bootstrap confidence values from 1000 iterations. Relative density (percent of individuals of each species in each collection) data were used to investigate changes in the pattern of distribution unbiased by the magnitude

98 of abundance in a given collection. Correlation between the abundance of the 25 most common species and aggregation age and sulfide concentration were determined by

Spearman rank correlation. A sequential Bonferroni correction for confidence level was used to account for the potentially confounding effects of multiple comparisons.

Results

Thirteen tubeworm aggregations were collected at the two study sites (Table 4-1).

They contained between 121 and 1635 tubeworms. The majority of the aggregations were dominated by L. luymesi, which comprised and average of 72.4% of the tubeworms.

However, one of the aggregations was almost entirely composed of L. luymesi (1d,

93.0%) and one contained a slight majority of S. jonesi (2c, 51.3%). Aggregation ages estimated by the population growth model ranged from 8 to 157 years. The majority of aggregations were on the younger end of this range with 7 of the 13 having an estimated age under 30 years and only one estimated to be over 75 years of age (1f).

The collections contained a total of 91 species, with individual aggregations containing between 13 and 47 species (Appendix). It is possible that there were 94 total species, as the taxonomic status of the hydroids collected in the study remain unresolved, but appear to consist of at least 4 species. Number of species of macrofauna associated with an aggregation was significantly correlated to habitat size (r = 0.685, p = 0.10).

Fairly high levels of diversity, as measured by the Shannon-Weaver diversity index (H’ =

1.39 to 2.78) and Pielou’s index of evenness (J = 0.423 to 0.833) were found in the tubeworm communities. Diversity was not linearly correlated to aggregation age. Lowest

99 diversity (H’ and J) was found in the 3 youngest aggregations (H’ = 1.39 to 1.60, J =

0.423 to 0.623), and the oldest aggregation (H’ = 1.37, J = 0.505). However, standardized

species richness (number of species per m2) was highest in the 2 youngest

aggregations (Table 4-1).

Table 4-1: Characteristics of collected tubeworm aggregations arranged by rank age. Aggregation name, as referred to in the body of the text, with numbers referring to site 1 (GC234) and site 2 (GC232) and superscripts referring to collection device (1 = Bushmaster Jr., 2 = Bushmaster Sr.). Included are depth of collection (m), abundance of the two vestimentiferan species, average length of L. luymesi (cm), average age (years) as estimated by the population growth model, tube surface area (m2), total number of species (richness), number of species per unit tube surface area, and diversity indices. Diversity is estimated by Shannon-Weaver diversity index (H’) and Pielou’s index of evenness (J). Standard deviations are in parentheses.

name depth L. luymesi S. jonesi avg length age surface area richness sp/m2 diversity evenness

1h1 528 601 117 12.5 (18.7) 8.0 (0.2) 0.338 19 56 1.597 0.542

1g1 539 92 29 14.6 (18.6) 8.2 (0.4) 0.060 13 217 1.598 0.623

2a2 569 421 222 42.7 (16.8) 21.2 (0.4) 1.884 27 14 1.394 0.423

1d1 540 581 44 29.2 (24.4) 21.6 (0.6) 1.031 27 26 2.431 0.738

2b1 569 150 67 55.1 (9.9) 24.0 (0.8) 0.899 22 24 2.572 0.832

1c1 538 216 137 47.9 (20.4) 26.0 (0.6) 0.985 28 28 2.776 0.833

1b2 534 873 350 49.9 (24.2) 29.4 (0.5) 3.771 40 11 2.912 0.789

2c1 571 74 78 59.3 (14.5) 40.1 (5.7) 0.591 19 32 1.974 0.670

1e2 527 1179 456 82.6 (15.1) 42.7 (0.5) 11.450 41 4 2.647 0.713

1j2 532 1443 147 97.3 (18.4) 59.7 (0.8) 5.460 24 4 2.524 0.794

1a2 534 460 266 77.1 (32.9) 61.4 (10.2) 4.316 43 10 3.118 0.829

1i2 535 1344 218 83.1 (28.4) 73.5 (12.5) 6.335 28 4 2.622 0.787

1f2 538 285 192 132.3 (35.7) 156.9 (36.2) 0.672 15 22 1.369 0.505 100 Endemic species dominate the majority of the communities collected. In terms of biomass (gAFDW per m2 tube surface area), the mussel Bathymodiolus childressi, the gastropod Bathynerita naticoidea, and the shrimp Alvinocaris stactophila were the most dominant species. Munidopsis sp. 1, Alvinocaris stactophila, and Provanna sculpta were the most abundant (number of individuals per m2), and were found in 12, 12, and 10 of the 13 collections, respectively. The only species which was found in all collections was the polynoid polychaete Harmothoe sp.

Similarity between communities can be assessed on a relative scale using multi- dimensional scaling (MDS) analysis (Fig. 4-2). The most similar communities were collected along with aggregations 1a and 1b (BC = 76.8), which were only meters from each other at site 1 (GC234). The next most similar were 2b and 1d (BC = 67.4), which were collected at 2 different sites but had very similar estimated ages (1d = 21.6, 2b =

24.0 yrs).

In the subset of aggregations for which sulfide data were available, similarity in age exhibited the highest correlation to community similarity (r = 0.525, p < 0.001, n =

10) of any of the variables tested (depth, longitude, sulfide, age). When all collections were entered into the analysis and sulfide removed as an environmental variable, similarity in age again provided the most highly significant correlation to community similarity (r = 0.583, p < 0.001, n = 13). The majority of the variability explained by aggregation age lay along the x-axis (x vs. log(age) r = 0.852, p < 0.001) in the MDS plot.

101

Figure 4-2: Multi-dimensional scaling (MDS) plot of community similarity. Placement of samples in 2-dimensions based on rank Bray-Curtis pair-wise similarity in community structure. Samples which appear closer together have relatively greater pair-wise similarity values, with a stress value of <0.1 indicating a reliable representation of community similarity in the ordination. Solid circles represent relative age of aggregation, with larger circles representing older aggregations.

After age, average sulfide concentration (r = 0.337, p = 0.024) and concentration at sediment level (r = 0.359, p = 0.015) exhibited the highest correlations to community similarity in the bushes for which sulfide data were available. There was a general trend of declining sulfide concentrations with age (Fig. 4-3), with average sulfide concentration

(r = -0.680, p = 0.030) and concentration at sediment height (r = -0.753, p = 0.012) exhibiting inverse relationships with age. However, other measures of sulfide concentration were not significantly correlated with age (sulfide presence r = -0.527 p =

0.118, plume r = -0.445 p = 0.197, mid r = -0.265 p = 0.460).

102

2.5

2

1.5 plume mid 1 sed

0.5 sulfide concentration (uM)

00 0 0 5 10 20 50 100 200 age (yrs)

Figure 4-3: Sulfide concentration in tubeworm aggregations. Age of aggregation as estimated by the population growth model shown along the x-axis in log scale. Sulfide concentration was measured in water samples taken within tubeworm aggregations at plume level, mid-length of tubeworms, and near the sediment-water interface. The symbol “0” represents no detectable sulfide in any sample taken at that aggregation. Error bars represent standard deviation.

The effect of distance was negligible in the MDS analysis including all collections

(r = 0.056, p = 0.715). This was emphasized by the tendency for collections from the two sites to cluster among each other (Fig. 4-2). Depth also had an insignificant correlation to community similarity (r = 0.011, p = 0.943), showing that the minor changes in depth in this study (44 m) had little effect on community structure.

The effects of aggregation age on community composition are most evident in the shift in trophic structure over time (Fig. 4-4). The proportion of biomass in the primary producer (r = -0.556, p = 0.049) and primary consumer (r = -0.571, p = 0.041) categories

103 significantly declined with age with secondary consumers demonstrating a concomitant rise in relative dominance (r = 0.877, p < 0.001). Higher level consumers showed no significant trend with age (r = 0.375, p = 0.206). The second youngest collection (1g) was the only one to have significant biomass (84%) in the form of the symbiotic mussel,

Bathymodiolus childressi. Primary consumers made up approximately 90% of the non- primary producer biomass in the two youngest collections (approximately 8 years of age).

The next 6 aggregations, ranging from 21 to 40 years of age, contained 50% to 68% of the associated biomass in the first trophic level. Aggregations in this stage contained the occasional, often solitary, higher predator (fishes or decapod crustaceans). Older aggregations (>40 years of age) contained a greater proportion of biomass in upper trophic levels, with less than 40% as primary consumers. In the oldest aggregation (157 years), secondary consumers comprised 93% of the biomass. Temporal changes in trophic structure were accompanied by significant declines in (log-transformed) overall biomass (r = -0.625, p = 0.022), density (r = -0.772, p = 0.002), and number of species per m2 (r = -0.662, p = 0.014) with age. The proportion of endemic species in the community also decreased in older aggregations (r = -0.681, p = 0.010).

104

1

0.9

0.8 0.7

0.6 PP PC 0.5 SC 0.4 HC 0.3 proportion of biomass 0.2

0.1

0 5 10 20 50 100 200 age (years)

Figure 4-4: Changes in trophic structure with aggregation age. Tubeworm aggregations are arranged along the x-axis in order of increasing age. Primary producers (pp), including symbiotic bivalves but not tubeworms, rapidly decline in abundance. Primary consumers (pc), mostly grazing gastropods and crustaceans, significantly decline in relative biomass over time. Secondary consumers (sc), including predatory polychaetes and crustaceans, significantly increase with aggregation age. Higher order consumers (hc) show no significant trend with aggregation age and represent the occasional intrusion by solitary fishes, seastars, or large crustaceans.

Consistent patterns of species co-occurrence gave rise to the observed trends in community structure. The most similar groupings of species involved endemic species which declined in abundance over time ( Fig. 4-5 ). It should be noted that the majority of correlations between species abundance and aggregation age or sulfide concentration were significant at the 0.05 level, but not at the Bonferroni-corrected level (p = 0.002).

One of the most highly supported associations was between the endemic shrimp

Alvinocaris stactophila and the endemic galatheid Munidopsis sp. 1 with a Bray-Curtis

(B-C) similarity of 76.2. Alvinocaris stactophila abundance significantly declined in

105 older aggregations (r = -0.825, p = 0.001) and increased with average sulfide concentration (r = 0.644, p = 0.044), but Munidopsis sp. 1 showed no significant trend with age or sulfide. The deposit-feeding sipunculan, Phascolosoma turnerae, declined over time (r = -0.760, p = 0.003) and clustered with the predatory polychaete Glycera tesselata and the nemertean. The methanotrophic mussel Bathymodiolus childressi clustered with the endemic gastropods Provanna sculpta and Bathynerita naticoides (B-C

= 74.7). B. childressi, P. sculpta, and B. naticoides all declined in abundance over time (r

= -0.812, p = 0.001; r = -0.755, p = 0.003; r = -0.637, p = 0.019 respectively) and increased in abundance with increased average sulfide concentration (r = 0.703, p =

0.023; r = 0.695, p = 0.026; r = 0.713, p = 0.021 repectively). The polynoid polychaetes

Branchinotogluma sp. and Harmothoe sp. often co-occurred and showed a trend of decreasing abundance in older aggregations (Branchinotogluma r = -0.733, p = 0.004;

Harmothoe r = -0.660, p = 0.014) and increased abundance in more sulfidic habitats

(Branchinotogluma r = -0.671, p = 0.034; Harmothoe r = -0.669, p = 0.034). These polychaetes were distributed similarly to the gastropod Cataegis meroglypta (B-C = 67.2) with declining abundance over time (r = -0.660, p = 0.014) and increasing abundance with sediment level sulfide (r = 0.740, p = 0.014).

106

Figure 4-5: Similarity in species distribution among aggregations. Cluster analysis (group average linkage) based on Bray-Curtis similarity of standardized density of species among aggregations. Trophic levels (tl) are as follows: (1) primary producer, (2) primary consumer, (3) secondary consumer, (4) higher consumer. Residency status (r) includes endemic species (e), non-endemic species (n), and species of unknown status (u). Correlation of species distribution with age (a) and sulfide concentration (s) shown as positive (+) or negative (-) with one symbol representing a significant correlation at 0.05 confidence level and two symbols denoting a significant correlation at Bonferroni- corrected confidence level (p = 0.002). 107 A number of species increased in abundance over time. The shrimp Periclimenes

(r = 0.569, p = 0.042) and the galatheid Munidopsis sp 2 (r = 0.637, p = 0.019) were more abundant in older aggregations and shared similar distributions (BC = 61.7). Other species with elevated abundance in older aggregations were the predatory or scavenging crustacean Rochina tanneri (r = 0.619, p = 0.024), and the predatory polychaete Eunice sp. (r = 0.569, p = 0.042). The rest of the common species showed no temporal pattern in abundance, and only clustered at low similarities and low bootstrap values with others.

Discussion

Vestimentiferan tubeworms profoundly affect the ecology of the ULS of the Gulf of Mexico by providing habitat for a diverse community of macrofauna. The community structure of the aggregations sampled demonstrates that there are predictable successional processes occurring within tubeworm-associated communities at these sites. As tubeworm aggregations age, trophic structure is altered with lower trophic levels declining in relative dominance while higher trophic levels increase. The proximate cause of these changes in community structure with aggregation age is the decline in sulfide concentration within the tubeworm generated habitat. Sulfide concentration, most significantly at sediment level, explains some of the variability in the MDS pattern of similarity in community structure, and a number of species showed a response in their distribution with sulfide concentration.

There were 94 species (not including the 2 species of tubeworms) collected along with the 13 aggregations in this study. A previous investigation of the community

108 structure of these habitats listed 65 species found in 7 aggregations (Bergquist et al.

2003), including 2 species of Alvinocaris which have since been determined to be 1 species (E.E.C., S.H. unpublished data). The species reported here increase the total number of species known from tubeworm aggregations on the upper Louisiana slope of the Gulf of Mexico to 102. Some of these species are not only new records for tubeworm aggregations, but are also new to science. Three of the gastropods, Gymnobella sp.,

Odostomia sp., and the limpet Iothia sp., as well as the ophiuroid are as yet undescribed new species which were not collected in previous studies at these sites.

These are fairly diverse communities, particularly for the relatively small habitat size investigated and considering that only species retained on a 2 mm sieve are considered in this analysis. The values of the diversity index (H’) determined here (1.39 to 2.78) are greater than mussel beds from the ULS (0.74 to 1.42, Bergquist et al. in press) and are comparable to the diversity of the fauna retained on a 250 µm mesh at other seep mussel beds from the Florida Escarpment and Blake Ridge (~2.5, Turnipseed et al 2003). The diversity of ULS tubeworm aggregations also appears to be greater than that found in Ridgeia piscesae aggregations on the Juan de Fuca ridge vent system (0.9 to

1.3, Govenar et al 2002), and that found in deep water Lophelia pertusa coral habitat from Norway (0.72 to 1.44, Mortensen et al 1995). The community collections presented here therefore represent some of the most diverse communities sampled within deep-sea biogenic structures.

The size of a collection was constrained by the size of the sampling apparatus, with the larger of the two having an open diameter of 1.5 m. At the two sites sampled here, most tubeworm aggregations were too large to sample and clusters of aggregations

109 forming uninterrupted fields covering hundreds of square meters were present. Since the number of species present was a direct function of the size of the habitat, it is likely that these larger habitat areas contain higher diversities of associated fauna. This is also predicted by the theory of island biogeography where larger habitat “islands” contain more diverse communities (MacArthur and Wilson 1963). According to this theory, increased diversity results not only from reduced competition for space but also the more complex habitat structure provided in larger islands.

Diversity of the community appears to increase initially and then decline in much older (>100 years) aggregations. It is expected that the overlap between the endemic and non-endemic communities in the mid-range of aggregation age should produce the highest diversity communities (Bergquist et al 2003). This was consistent with the diversity (H’) and evenness (J) indices, but not diversity measured as species richness.

This is due to the fact that the youngest communities are comprised of a few dominant species and many rare species, resulting in the discrepancy between species richness and diversity indices.

Distance between collections had little influence on the distribution of species.

The collections from the two study sites (GC234 and GC232) did not segregate in 2 dimensional space in the MDS plot. In fact, collections from two different sites (1d and

2b) had the second highest Bray-Curtis similarity (67.4). Since the two sites lie less than

10 km apart, the lack of a barrier to dispersal, or even migration for the more mobile species, may explain the high degree of overlap in community composition. In this situation, the tubeworm aggregations do not necessarily fit the theory of island biogeography. A high proportion of the community is comprised by non-endemic species

110 that are found on the surrounding soft bottom in addition to occupying tubeworm aggregations. In this way, the aggregations are interconnected to a greater extent than entirely isolated islands allowing them to exist in a meta-community context.

The two most similar communities (1a and 1b, BC = 76.8) were collected less than 2 m apart, suggesting that there is some effect of distance on very small spatial scales. The mobility of species comprising the upper trophic levels in conjunction with the proximity of adjacent habitats could lead to the dominance of upper trophic levels in the community structure of older aggregations. Secondary consumers and scavengers are likely taking advantage of the shelter afforded them by the relatively innocuous biogenic habitat to forage in surrounding mussel beds and younger tubeworm aggregations containing higher sulfide levels and prey abundance (Bergquist et al. in press). This is shown in the occasional intrusion of upper level predators (fishes and large crustaceans) into younger aggregations (Fig. 4-4). The relatively high biomass represented by a single individual of these species gives rise to the sporadic pattern of abundance of this group. It should be noted that there is a confounding factor in the analysis of these two particular aggregations (1a and 1b). A few days prior to collection, the top 5-20 cm of sediment surrounding the tubeworms was suctioned off in an attempt to quantify the extent of the

“rhizosphere” (Chapter 3) of one of the aggregations. Because this did not appear to significantly alter the community composition of the associated fauna, the data obtained on the two submersible dives required to collect these aggregations were included in the study.

The successional pattern of change in community structure along with declines in overall biomass and density of fauna with age have been hypothesized to be a result of

111 lower sulfide concentrations in older aggregations (Bergquist et al 2003a). This follows the logic that primary productivity within a tubeworm aggregation should decline as sulfide concentration in the epibenthic water layer lessens. This was seen in the correlation of community similarity with average and sediment level sulfide concentration. Since the many of the organisms inhabiting the aggregations were epibenthic or infaunal species, concentration near the sediment surface may influence their distribution significantly. The combination of lower primary productivity and changes in environmental toxicity are likely contributors to the observed changes in community structure.

Significant changes in community structure with sulfide concentration have been shown in a variety of habitats. Three distinct community types are associated with different sulfide regimes found overlying gas hydrates on the Cascadia convergent margin (Sahling et al. 2002). The composition of infaunal communities at northern

California methane seeps are directly related to interstitial sulfide concentrations (Levin et al. 2003). Flow rates of seeping fluids affect the chemosynthetic community structure around the periphery of mud volcanoes near the Barbados accretionary prism (Olu et al

1997).

Though aggregation age explained a greater proportion of the variability in community structure, sulfide concentration is the likely proximal cause of the changes in community structure. The vagaries of sampling sulfide in the 0.2 to 2.0 µM range may have contributed to the variability associated with the sulfide measurements. The range associated with the replicate samples taken around individual bushes (Fig. 4-3) suggests that there is great temporal and/or spatial variability in water column sulfide

112 concentration. Similar variability in epibenthic and interstitial sulfide concentration has been previously noted at these sites (Bergquist et al 2003b). The subtleties of changing sulfide concentration over the course of decades could be obscured by this variability.

The detection of significant trends in the sulfide concentration sampled here even in the presence of the high variability between replicates belies the great impact of this factor on the community structure of these habitats.

Changes in sulfide concentration over time may be manifest in patterns of vestimentiferan population structure which reflect their exposure to sulfide over longer temporal scales than discreet samples can detect. Recruitment patterns are influenced by sulfide concentration in that vestimentiferan larvae require sulfide to metamorphose and grow (Tyler and Young 1999). The estimated upper limit for the settlement period of L. luymesi in a single aggregation is 40 to 60 years (Bergquist et al 2000, Chapter 2). Of the

7 aggregations under 40 years of age in this study, only 2b did not exhibit active recruitment. The only aggregation over 40 years of age to contain recent recruitment was the 61 year old aggregation (1a). While micromolar levels of sulfide were detected among the tubes of this aggregation, no sulfide was detectable in the 3 other aggregations of comparable age or older. This evidence suggests that sulfide should decline to undetectable levels within this time period, coinciding with the cessation of settlement and decline in primary consumer biomass observed.

Previous studies have suggested that L. luymesi may alter its environment by reducing sulfide flux from the sediment (Chapter 3). This tubeworm species obtains sulfide through posterior extensions of its body, the “root.” (Julian et al. 1999, Freytag et al. 2001). High uptake rates, exceeding 30 mmol·hr-1 in moderate sized aggregations

113 (Chapter 2), will serve to deplete sulfide as it is carried upwards by seeping fluids. The roots may physically impede seepage by growing down through existing seepage conduits (Bergquist et al 2003a). Theoretical studies have demonstrated the high probability that L. luymesi further affects sulfide flux by releasing sulfate into deeper sediment layers (Chapter 3). If suitable electron donors are available, sulfate will be reduced to sulfide in deeper sediment layers. By shifting the site of sulfate reduction, gaseous and liquid hydrocarbons seeping upwards with pore fluids will be depleted at greater sediment depths. The amount of methane escaping the sediments into the epibenthic layers would be greatly reduced, precluding the growth of Bathymodiolus childressi in older aggregations. This is consistent with B. childressi’s significant decline in abundance with time in vestimentiferan aggregations. Additional restrictions of seepage pathways will occur through carbonate precipitation, a byproduct of anaerobic methane oxidation and hydrocarbon degradation (Boetius et al. 2000, Formolo et al.

2004).

While sulfide concentration is the most likely proximal cause of the successional pattern observed, similarity in community structure between aggregations is best explained by the relative age of the aggregation. Young aggregations consist of primarily endemic species of lower trophic levels. The mussel with methanotrophic symbionts,

Bathymodiolus childressi, and the commonly co-occurring gastropod Bathynerita naticoides, appear to decline in abundance within the first 10 to15 years. It should be noted that the dominance of these species in the second youngest collection (1g) is likely due to the fact that it was collected adjacent to a Bathymodiolus mussel bed, a common location for young tubeworm aggregations. This does not completely explain its

114 similarity to the youngest aggregation (1h) since this collection was not adjacent to a mussel bed and Bathymodiolus was far less abundant (6th in rank density and rank biomass). It has been suggested that mussel beds represent an earlier successional stage to tubeworm aggregations (Bergquist et al in press). This hypothesis fits with the pattern of community similarity shown here and is supported by the common finding of encrusted mussel shells in the authigenic carbonate found at the base of many the tubeworm aggregations (pers. obs.). While primary producer species still appear in older aggregations, these are more commonly burrowing bivalve clams with chemoautotrophic symbionts. By mining sulfide from within the interstices of shallow sediments, these autotrophs avoid the limitation of sulfide in the water column.

Over the course of the next 20 to 30 years, decreasing nutritional resources and increasing predation lead to declines in the biomass of grazers and primary consumers.

Local primary productivity in the form of bacterial biomass should decline with time, inhibiting the growth and reproduction of species in lower trophic levels. Reduced sulfide levels in the epibenthic layer would facilitate colonization by species in higher trophic levels with lower sulfide tolerances. The species which appear to be most adversely affected by these environmental changes are the endemic shrimp Alvinocaris stactophila, the symbiotic mussel (B. childressi) and associated gastropods (B. naticoidea, and P. sculpta), the endemic polynoid ploychaetes, the gastropod C. meroglypta, and the deposit-feeding sipunculan (Phascolosoma turnerae) (Fig. 4-5). All of these species exhibited significant declines in abundance with time. The symbiotic mussel has direct ties to elevated environmental concentration of reduced chemicals, as these chemicals serve as their ultimate source of nutrition. B. naticoides is the most common associate of

115 B. childressi within mussel beds (Bergquist et al. in press), and contains fungal symbionts which may assist it in sulfide detoxification (Zande 1999). The endemic grazers and micro-predators are able to utilize the increased local productivity, perhaps due to their tolerance to sulfide toxicity. C. meroglypta and P. turnerae are non-endemic species which show increased abundance at seep sites (Carney 1994) and may also have elevated sulfide tolerances which allow them to exploit chemosynthetic production. Once sulfide levels decline, the tolerance of these species to sulfide would cease to afford them a competitive advantage.

The biomass of primary consumers continues to decline until it is exceeded by the biomass of secondary consumers in aggregations of 40 to 60 years of age. The modification of the tubeworm habitat facilitates colonization by those species with lower sulfide tolerances. The most significant beneficiaries of lowered environmental toxicity are predatory and scavenging species. The galatheid Munidopsis sp 2 and the species it shared the most similar distribution with, the shrimp Periclimenes sp., are likely competing with 2 other species, Munidopsis sp. 1 and the bresilid shrimp Alvinocaris stactophila. The distribution of Munidopsis sp. 1 is correlated to A. stactophila which significantly declines in abundance over time, suggesting that in the absence of sulfide,

Munidopsis sp. 2 and Periclimenes have the competitive advantage. It is possible that this is a general trend of species with high sulfide tolerances being replaced by species with better resistance to predation.

Other species that occur more commonly in older aggregations inhabit the upper portions of aggregations away from the sediment surface and the source of sulfide. These include the mobile crab R. crassa and the polychaete Eunice sp. which forms tubes on the

116 anterior portions of the vestimentiferan tubes. Concomitantly, hydrozoans and serpulids

(not included in quantitative analyses due to the difficulty of quantifying colonial organisms) exhibit increased frequency in older aggregations, being present on the majority of tubeworm tubes in older aggregations (E. Cordes, unpublished data). This stage appears to persist until primary level consumers are nearly absent in aggregations approaching 200 years of age. It is unknown how long this final stage of succession may endure in an individual tubeworm aggregation, as the upper limit of L. luymesi longevity is unknown. The paucity of community structure and growth data is mainly due to the constraints imposed on collecting large, old tubeworm aggregations by an 8 m long submersible with a 450 kg payload using extant collection equipment.

Upper trophic levels rarely dominate a community in terms of biomass. This is somewhat deceiving, as the biomass of the vestimentiferans in the aggregations, which accounts for over 99% of the total biomass in each collection, was not included in the trophic analysis. However, evidence from stable isotopic and anti-predatory defense compound studies indicates that there is little or no predation on the tubeworms at these sites (MacAvoy et al. in press, Kicklighter et al., in press). Therefore, the high biomass represented by the tubeworms does not appear to be the energetic input at the base of the food web. This is analogous to the role of trees in a forest where the vast majority of standing biomass is contained in the labile materials of the trunk and branches.

L. luymesi maintains positive interactions with multiple species not only by creating habitat, but also by modifying the abiotic conditions within its habitat. While certain endemic species are inhibited by these modifications, the colonization of other endemic and non-endemic species is facilitated. Since it is an extremely long-lived

117 organism, these effects persist on temporal scales measured in centuries. The abundance of L. luymesi and the size of the physical structure it creates make it the most significant foundation species at hydrocarbon seeps on the upper Louisiana slope. Further, the spatial and temporal scale on which its habitat persists rank it near the deep-water coral reefs of

Lophelia pertusa (Mortensen et al. 1995) as one of the most significant foundation species in the deep-sea as a whole.

In summary, tubeworm aggregations exhibit predictable changes in community structure over time. The pattern of community succession shown here validates the model put forth by Bergquist et al. (2003a). In the first 10-15 years, the community contains high biomass dominated by lower trophic levels relying on local chemosynthetic productivity. Over the next 30-50 years, primary consumers decrease in biomass with secondary consumers eventually exceeding the biomass of primary consumers. While biomass steadily declines in this stage, aggregations contain higher levels of diversity by harboring a mixture of endemic and non-endemic species. In aggregations approaching

200 years of age, primary consumers are nearly absent from the community. The proximal cause of these successional changes is sulfide concentration within the aggregations, most significantly at sediment level. The role of L. luymesi in creating the physical structure that harbors these communities and the potential role of L. luymesi in altering the sulfide distribution within the habitat it creates demonstrate the profound influence it has on the community ecology over large areas of the continental slope of the

Gulf of Mexico.

118 Notes

Stephane Hourdez, Ben Predmore, Meredith Redding, and Chuck Fisher will be coauthors on the eventual publication resulting from this Chapter and have edited previous versions of this manuscript. Stephane Hourdez assisted with all aspects of data collection at sea and in the laboratory. Ben Predmore collected the majority of the sulfide concentration data included here. Ben Predmore and Meredith Redding are responsible for the majority of the tubeworm population data included here. Chuck Fisher provided logistical support and collected most of the samples included in the analysis. Chris Jones is responsible for collecting some of the sulfide concentration data included in this study.

Julie Barsic, Brian Tiegs, and Nicole Iacchei all contributed to the enumeration and measurement of tubeworms and associated fauna. I would also like to acknowledge the assistance of Breea Govenar, Sharmishtha Dattagupta, Guy Telesnicki, Mike McGinley,

Liz Podowski, and the Craig Young lab for their assistance with collections at sea.

Chapter 5

Impact of an Ecosystem Engineering Tubeworm on the Habitat Characteristics and Community Structure of Gulf of Mexico Hydrocarbon Seeps

Abstract

Lamellibrachia luymesi is a long-lived vestimentiferan tubeworm inhabiting deep- sea hydrocarbon seeps in the Gulf of Mexico. It has been hypothesized that this species significantly alters its habitat through uptake of hydrogen sulfide from seep sediments and release of sulfate into deeper sediment layers. Here, we examine the structure of communities associated with 18 tubeworm aggregations at 7 sites on the upper Louisiana slope. We use growth models of L. luymesi and S. jonesi to place these communities in a temporal sequence. Changes in epibenthic sulfide distribution with estimated aggregation age produce substantial changes in the community structure at these sites. The temporal component of these changes in community structure follows a different trajectory in tubeworm aggregations comprised of different proportions of L. luymesi. While sulfide generally declines over time, the proportion of L. luymesi within a tubeworm aggregation explains a significant part of the variability in epibenthic sulfide concentration. Together, this evidence demonstrates the profound influence of L. luymesi on its habitat and its role as an ecosystem engineer.

120 Introduction

Ecosystem engineering theory pertains to species that have a disproportionate effect on their environment, shaping the habitat in which they live (Jones et al., 1994).

Though the influence of organisms on the characteristics of their environment has long been recognized (Redfield 1958), it is only recently that this influence has been formalized into ecological theory (Jones et al., 1994, Bruno et al., 2004). While this theory has received increasing attention in recent years, direct evidence of the impact of habitat alteration by a single species on the surrounding community remains sparse (Lill and Marquis 2003).

The majority of ecosystem engineering theory is derived from examples in the terrestrial environment. The most well known example is the alteration of limnic systems by the ecosystem engineering beaver (Naiman et al. 1986), leading to increases in landscape scale diversity (Wright et al. 2002). Other examples include shelter-building caterpillars increasing the diversity of insects inhabiting white oak (Lill and Marquis

2003), prairie dog burrows altering soil nutrients (Whicker and Detling 1988), and elephant grazing and physical disturbance altering plant communities (Naiman 1988), among many others (see review in Jones et al. 1994). More recently, ecosystem engineering theory has been expanded to include marine systems. Examples include the complex habitat created by the solitary ascidian Pyura praeputialis which supports highly diverse communities in the intertidal zone (Cartilla et al. 2004), burrows of the tilefish providing habitat for other fishes on the continental shelf (Coleman and Williams 2002),

121 and the varied mechanisms by which molluscan shell production alters the marine environment (Gutierrez et al., 2003).

Changes in community structure resulting from changes in the abiotic environment of chemosynthetic ecosystems are well known (Johnson et al. 1988, Hessler et al. 1988, Luther et al. 2001, Sahling et al. 2002). At hydrothermal vents changes in community structure occur along gradients of fluid flux as a result of differing temperature and sulfide concentration (Mullineaux et al. 2003). In seep systems, substantial changes in community composition can occur across sulfide gradients in the micromolar range (Bergquist et al. in press), with the majority of infaunal species avoiding habitats with sulfide concentrations in excess of 1 mM (Levin et al. 2003).

A variety of symbiotic organisms create biogenic habitat at hydrothermal vents and cold seeps, including bathymodiolin mussels, vesicomyid and lucinid clams, and vestimentiferan tubeworms (see review in Childress and Fisher 1991). Tubeworms and clams contain internal sulfide-oxidizing chemoautotrophic symbionts, and mussels contain methanotrophic, chemosynthetic, or both types of symbionts (see review in

Nelson and Fisher 1995). They can create complex physical structures which are colonized by a diverse group of fauna (Carney 1994, VanDover 2002, Urcuyo et al. 2002,

Bergquist et al. 2003a). Lamellibrachia luymesi is the most abundant vestimentiferan on the upper Louisiana slope (ULS) of the Gulf of Mexico (MacDonald 1990). It forms aggregations of hundreds to thousands of individuals, normally including one or two other species of vestimentiferan, Seepiophila jonesi and/or an undescribed tubeworm of the Escarpia-group. Seepiophila jonesi is the more common of the two other species, normally composing approximately 30% of L. luymesi dominated aggregations

122 (Bergquist et al. 2003a) and occasionally dominating aggregations with only a few L. luymesi. The hypothesized ability of L. luymesi to influence the levels of sulfide in its habitat suggests that this tubeworm may have an impact on the communities inhabiting the structure it creates.

Hydrothermal vent vestimentiferans like the well studied Riftia pachyptila obtain sulfide from the water column across their highly vascularized plumes (Childress and

Fisher 1991). In contrast, L. luymesi is capable of utilizing a posterior extension of its body, the “root” to acquire sulfide from seep sediments (Julian et al. 1999, Freytag et al.

2001). The high biomass contained in even moderate sized aggregations leads to very high sulfide uptake rates from seep sediments (Chapter 2). Theoretical models suggest that in order to meet this requirement, L. luymesi releases the sulfate generated by its symbionts back into sediment layers below the normal depth of sulfate penetration

(Chapter 3). This would supplement the amount of sulfide produced through anaerobic methane oxidation, organic matter decomposition, and hydrocarbon degradation.

Increased integrated rates of hydrocarbon oxidation would also lead to higher rates of carbonate deposition in deeper sediments. The depletion of sulfide, decreased methane concentrations in surface sediments, and enhanced carbonate precipitation are all processes by which L. luymesi may alter the habitat which it creates.

In this study, we investigate the structure of tubeworm-associated communities from 7 hydrocarbon seep sites in the Gulf of Mexico, including 5 sites that have not been previously quantitatively sampled. We examine the community structure data with respect to the age of the aggregation, the concentration of sulfide in the tubeworm habitat, and the proportion of L. luymesi in the aggregation. In order to elucidate the causes of

123 observed changes in community structure, we examine the relationship between sulfide concentration and age of aggregations of varying tubeworm species composition. We discuss these findings with respect to predictions made by previous modeling studies that

L. luymesi exerts a significant influence on environmental sulfide concentration at hydrocarbon seeps. These data provide evidence for the ecosystem engineering role of L. luymesi in determining the composition of communities associated with tubeworm aggregations at hydrocarbon seeps in the Gulf of Mexico.

Methods

A total of 18 vestimentiferan aggregations were collected on 4 cruises aboard the

R/V Seward Johnson II from 2002 to 2004. Collections were made at 7 different sites in the northern Gulf of Mexico (Fig.Figure 5-1). The first 2 sites lie along the same fault scarp in the Green Canyon (GC) 234 (site 1; 27:44.8N, 91:13.3W, 530 m depth) and

GC232 (site 2; 27:44.5N, 91:19.1W, 570 m) lease blocks of the ULS. The sites were approximately 10 km apart and ranged from 528 to 571 m depth. Community structure data from the collections obtained at these sites (1a, d, f-j, 2a-c) were examined previously in Chapter 4 (see Appendix). Just north of these sites, in GC233 (site 3;

27:43.4N, 91:16.8W, 650 m), lies a basin filled with a hypersaline brine. The borders of this brine pool are dominated by the mussel Bathymodiolis childressi with small aggregations of tubeworms located nearby. Site 4 was in GC354 (27:35.9N, 91:49.4W,

560 m), approximately 60 km to the west of these sites on the flanks of a large mound.

Site 5 is in the Mississippi Canyon (MC) 885 lease block approximately 150 km to the

124 east of GC234 (28:03.7N, 89:42.6W, 620 m). This site was characterized by extensive areas of seepage associated with large carbonate outcrops. Site 6 lies 200 km to the north- east of MC in the Viosca Knoll (VK) 826 lease block (29:09.3N, 88:01.4W, 540 m) on the crest of a large mound with abundant seepage and carbonate. Site 7 was in the Garden

Banks (GB) 543 lease block located approximately 200 km to the west of GC234

(27:26.9N, 93:11.1W, 550 m). This area of Garden Banks is the site of multiple areas of localized brine seepage with high concentrations of lighter hydrocarbons, primarily methane.

Figure 5-1: Map of collection locations in the northern Gulf of Mexico.

Intact vestimentiferan aggregations were obtained using the Johnson Sea-Link I and II submersibles with the Bushmaster Jr. and Bushmaster Sr. collection devices. The

Bushmaster collection devices are hydraulically actuated nets lined with a 63µm mesh.

125 Community samples were sieved through a 2 mm mesh and sorted on board ship to lowest possible taxonomic level. All associated fauna were preserved and transported back to Penn State for final determination of taxonomic status. E.E.C. and S.H. identified all polychaetes. Primary identification of other groups was carried out and specimens sent to experts for further identification or verification. All individuals were weighed and enumerated for subsequent analyses. Tubeworms were counted and measured on board, time permitting (see Chapter 4 for complete methodology).

Aggregations were assigned a relative age based on the growth rates of the two tubeworm species. Growth data were obtained by staining whole aggregations of tubeworms with a chitin stain (acid blue #148). The Bushmaster Jr. collection device was lined with a layer of plastic and attached to a bladder on the top of the submersible. The

Bushmaster was constricted around an aggregation and the stain pumped into the device.

The tubeworms were exposed to the stain for approximately 5 minutes, with additional stain added occasionally to maintain super-saturation. Two aggregations were stained at both GC234 and GC232 in June 2002 and collected in August 2003. Two additional aggregations were stained in August 2003 and collected in June 2004. Growth data from these 6 aggregations were combined with the data from Bergquist et al. (2000) to model

L. luymesi and S. jonesi growth.

Growth rate was determined by least-square regression (JMP®, SAS Institute,

1997). Growth data for L. luymesi were fitted to the function: g = aε −bL (1) where g = growth rate in cm·yr-1, a = 4.903, b = 0.01499, and L is tube length in cm. To estimate the variability around growth rate, equation (1) was fitted to the residuals of the

126 previous function where a = 2.033 and b = 0.00647. Growth rate of S. jonesi did not significantly vary with size. However, the proportion of individuals which exhibited positive growth did vary significantly with size. Therefore, S. jonesi growth was modeled using a set of two equations. The first describes the size specific probability that S. jonesi grows based on the frequency of positive growth in 10 cm size classes: p(0) = aln(L) + b (2) where p(0) is the probability of no growth, a = 0.3074, b = -0.2865, and L = tube length.

If growth is positive, then the magnitude of growth is determined by the average annual growth (mean = 2.13 cm, sd = 1.55 cm). Average age at size was determined for each species by 1000 iterations of the two growth models.

Water samples were taken among the tubes of the tubeworms at plume height, mid-level within an aggregation, and at the base of the aggregation near the sediment- water interface. The samples were drawn through 0.5 mm diameter PEEk tubing into a gas-tight syringe in the back of the submersible. Once the submersible is retrieved, hydrogen sulfide concentration is determined using a modification of an enzymatic assay

(Freytag et al. 2001, Bergquist et al. 2003b) according to the methodology cited in

Chapter 4.

The relationship between sulfide concentration and age and proportion of L. luymesi in each aggregation was determined with regression analyses. Sulfide concentration was entered into the model as average concentration, and concentration at plume, mid, and sediment-level. All sample values under 0.2 µM were not significantly different from 0 (confidence interval of regression overlapped 0), and were therefore treated as 0 in the analysis. Exact values for concentrations determined above 2 µM could

127 not be given with confidence since they were outside of the range of the standard curve, and were therefore treated as a value of 2 µM in the analysis. A measurement in excess of

2 µm occurred in only 4 of 126 samples. Sulfide concentration was log-transformed

(Log(x+1)) to account for its skewed distribution. Rank age of the aggregation was entered into the model along with log-transformed proportion of L. luymesi.

Similarity in community structure between aggregations was examined using multidimensional scaling (MDS) using PRIMER software (PRIMER-E ltd.). Species abundance was standardized to tubeworm surface area to provide a measure of species density irrespective of collection size. Density data were 4th root transformed to reduce the bias towards the most dominant species while maintaining the information provided by the more rare species (Clarke and Warwick 2001). Bray-Curtis similarity was determined using the following function:

⎛ p ⎞ ⎜ ∑ yij − yik ⎟ Sjk = 100⎜1− i=1 ⎟ (3) ⎜ p ⎟ ⎜ ∑()yij + yik ⎟ ⎝ i=1 ⎠ where yij is the abundance of the ith species in the jth sample and p is the total number of species. MDS is an iterative procedure that minimizes the difference between ranked

Bray-Curtis (BC) similarity values and 2-dimensional distance between each pair of samples. Following the MDS analysis, a modified BIO-ENV procedure (Clarke and

Warwick 2001) was used to examine potential explanatory variable on the pattern of similarity observed. Spearman rank correlation between Bray-Curtis similarity in community structure and Euclidean distance in habitat characteristics was used to determine which environmental factors best explained patterns in community similarity.

128 Variables tested included aggregation age, sulfide abundance, proportion of L. luymesi in the aggregation, distance (Euclidean distance in latitude and longitude combined), and depth.

Results

Seepiophila jonesi shows extremely slow growth rates (Fig.Figure 5-2a). The majority of animals collected (59.6%) had no appreciable growth over the year following in situ staining. Larger individuals (>30 cm) showed lower frequency of positive growth

(21.9%) than small individuals (44.5%). Based on these data, the growth model predicted an age of 73 years for a 50 cm animal. The largest S. jonesi collected, 111.1 cm, is predicted to be 299 years old by the model. While very few tubeworms of this size have been collected, this suggests that S. jonesi individuals attain ages comparable to L. luymesi. Age of aggregations varied between 2 and 299 years. The age estimates derived from the growth rates of L. luymesi and S. jonesi were highly correlated (r = 0.890, p <

0.001) (Fig. 5-2b).

129

5-2a

250

200

150 L. luymesi S. jonesi 100 length (cm) length

50

0 0 50 100 150 200 time (years)

5-2b

1000

100 S. jonesi

10

1 1 10 100 L. luymesi

Figure 5-2: Tubeworm growth models and estimated ages. 2a. Growth model of Lamellibrachia luymesi (top) and Seepiophila jonesi (bottom). Average, maximum and minimum length estimated by 1000 iterations of the two growth models. L. luymesi growth approximated from the growth model presented in Cordes et al (submitted). S. jonesi growth estimated from a two-stage model calculating probability of positive growth and magnitude of growth. 2b. Estimated ages of aggregations. Estimated age shown (log scale) with 1:1 trend line for reference. Ages estimated from Lamellibrachia luymesi and Seepiophila jonesi growth models based on average length of tubeworms in each aggregation. 130 Age of the aggregation explained a significant proportion of the variability in sulfide concentration (Fig. 5-3). The highest single value recorded was 4.3 µM

(aggregation 7b at sediment level), though this is reported without confidence as it exceeded the range of the standard curve (2 µM) in the assay. Average sulfide concentration significantly declines over time (F = 6.69, p = 0.020, r2 = 25.1) (Fig. 5-4a).

When the proportion of L. luymesi in the aggregation is added to the regression, the amount of variability in sulfide concentration explained by the model greatly improves (F

= 7.70, p = 0.005, r2 = 44.1) (Fig. 5-4b). The proportion of L .luymesi in the aggregation was not significantly related to average sulfide concentration ([H2S] = 0.217 – 0.578logL,

F = 3.22, p = 0.092, r2 = 11.5), though aggregations with higher proportions of L. luymesi tended to have lower sulfide values. Sulfide concentration measured at mid-aggregation level and at plume height did not significantly change with time, and variability in concentration between aggregations and among replicates within an aggregation was relatively high (Fig. 5-3).

131

2.5

2

1.5 plume mid 1 sed

sulfide concentration (uM) 0.5

0 1h 1g 4a 3b 1d 2a 5a 6b 7b 3c 6a 2b 2c 4b 1i 1a 1j 1f

Figure 5-3: Sulfide concentration within tubeworm aggregations. Hydrogen sulfide concentration measured at each of three heights among the tubes of the aggregations. Aggregations are arranged along the x-axis in order of increasing rank age. Error bars represent standard deviations.

5-4a 5-4b

0.3 0.3 0.25 0.25 0.2 0.2 0.15 0.15 0.1 0.1 0.05 0.05 0 0 predicted concentration -0.05 predicted concentration -0.05 00.10.20.3 00.10.20.3 actual concentration actual concentration

Figure 5-4: Correlations between predicted and measured hydrogen sulfide concentration. Measured sulfide concentration shown as log-transformed sediment level sulfide. Regression model in 5-4a includes aggregation age as a predictor (F = 6.69, p = 0.020, r2 = 25.1). Model shown in 5-4b includes aggregation age and proportion of L. luymesi (F = 7.70, p = 0.005, r2 = 44.1).

132 In the 18 tubeworm collections made, there were a total of 5570 individuals, representing 90 species (Appendix). The most abundant species in the collections were

Bathynerita naticoidea, Provanna sculpta, and Alvinocaris stactophila, all species which are endemic to hydrocarbon seeps. Endemic species comprised 7 of the 10 most abundant species, with 2 additional species having unresolved taxonomic and residential status.

Non-endemic species are present in lower numbers, with the sipunculan Phascolosoma turnerae, and the gastropods Eosipho canetae and Cantrainea macleani dominating this portion of the community.

Species overlap between communities was high. The most similar communities

(1g and 1h; 6a and 6b; 1d, 2b and 7b) had Bray-Curtis (BC) similarity values above 60%

(Fig. 5-5). The two most similar communities (6a and 6b, BC = 68.6) were both collected from Vioska Knoll, a site well removed from all of the other sites sampled in this study

(Fig. 1). These two aggregations were also of very similar age (rank 8 and 11) and contained similar proportions of L. luymesi (54.4% and 60.4%). Aggregations 1g and 1h were the two youngest aggregations in the study and contained high densities of the endemic species Bathymodiolus childressi and Bathynerita naticoidea. These aggregations were collected from the same site (GC234) and were composed of similar proportions of L. luymesi (1g: 76.0%, 1h: 83.7%).

133

Figure 5-5: Multidimensional scaling plot of community similarity. Species abundances were standardized to tubeworm surface area and fourth-root transformed. Similarity of communities calculated using the Bray-Curtis index. Collections are arranged in 2- dimensional space based on the rank similarity of communities. Stress value (0.16) indicates an adequate fit of the similarity matrix to the ordination. As a check on the validity of the ordination, aggregations that exhibit Bray-Curtis similarity values in excess of 55% are indicated (circles).

Aggregations 1d, 2b, and 7b contained high numbers of the bresilid shrimp

Alvinocaris stactophila, the galatheid Munidopsis sp 1, and the gastropod Provanna sculpta. These are all endemic species with presumably high sulfide tolerances, but reside in higher trophic levels than the most common species in the two youngest aggregations.

While these three aggregations harbor similar communities, they are different with respect to site (1d: GC234, 2b: GC232, 7b: GB535), age (1d: age rank 5, 2b: age rank 12,

7b: age rank 9), and proportion of L. luymesi (1d: 93.0%, 2b: 69.1%, 7b: 5.4%).

However, there were similarities in sulfide concentration among these three aggregations.

134 Aggregation 1d had relatively low sulfide concentration at sediment level (0.44 µM) for a young aggregation. Aggregation 2b had fairly low levels of sulfide, but sulfide was detected in one of the samples taken among the anterior end of the tubes. Aggregation 7b had among the highest sulfide concentrations of any aggregation with the second highest average concentration (0.814 µM), and a high proportion of samples containing detectable sulfide (4 of 6).

The community composition of tubeworm associated fauna is determined by a combination of aggregation age and epibenthic habitat chemistry. Similarity in aggregation age was significantly correlated to community similarity (r = 0.389, p<0.001), though the proportion of variation explained by this correlation was low.

Average sulfide concentration was also significantly correlated to community similarity

(r = 0.176, p = 0.030). The proportion of L. luymesi in the aggregation was not directly correlated to community similarity (r = -0.096, p = 0.239). The other environmental factors tested were insignificantly correlated to community similarity (depth: r = 0.048, p

= 0.556; distance: r = 0.015, p = 0.855).

Discussion

The results of this study demonstrate the profound effect L. luymesi has on its local environment. The structure of the macrofaunal communities associated with tubeworm aggregations is most closely related to the age of the aggregation, though some of the most similar communities were collected in aggregations of very different ages.

The likely proximal cause of these changes is the concentration of sulfide among the

135 tubeworm tubes (Bergquist et al. 2003b, Chapter 4). Specific patterns of sulfide concentration are different in aggregations composed of different species of vestimentiferans, with the relative abundance of L. luymesi significantly related to the sulfide concentration found in the aggregation. If capable of decreasing the level of sulfide in the epibenthic habitat, as predicted by previous modeling studies (Chapter 2, 3),

L. luymesi would act as both an autogenic and allogenic engineer by altering the quantity and distribution of sulfide within the physical structure it provides. We speculate that it is the process of amelioration of environmental toxicity by L. luymesi that makes its habitat accessible to a variety of species which would not normally be able to colonize seeps.

Aggregation age estimates were used to assess temporal changes in the environmental and biological characteristics of the aggregations collected. Estimates of age derived from the growth models for the 2 species were highly correlated. The ages of large S. jonesi individuals are on the same order of magnitude as the ages estimated for the oldest L. luymesi individuals. This finding places S. jonesi in the same category as the longest-lived non-colonial animal known, L. luymesi.

Since growth rates of tubeworms at sites other than sites 1 and 2 were not available, it is uncertain whether age estimates are consistent in a variety of habitats.

Existing geochemical data suggests that the conditions at the various sites are comparable. This is corroborated in the biological data as they harbor similar chemosynthetic faunas. However, the magnitude and variability of sulfide concentration available to the tubeworms may vary at the different sites, resulting in different growth rates. Therefore, the ages assigned in this study were used merely to rank the aggregations in terms of age rather than provide a quantitative estimate. Further in situ

136 staining of tubeworms at a variety of sites is required to resolve this potential source of error.

Hydrogen sulfide concentration varied both with aggregation age and by the proportion of L. luymesi in the aggregation. A decrease in sulfide concentration with aggregation age has been previously observed in L. luymesi dominated aggregations from sites 1 and 2 (Bergquist et al. 2003b, Chapter 4), as well as those from GC185 also known as Bush Hill (Bergquist et al. 2003b). Temporal trends in epibenthic sulfide abundance are related to the natural evolution of a seepage source with vigorous seepage initially in the form of active methane bubble streams (MacDonald et al. 1990a) followed by occlusion of shallow subsurface seepage conduits (Sassen et al. 1994, Aharon and Fu

2000) and reductions in organic materials fueling sulfate reduction in shallow sediments

(Sassen et al. 1999, Chapter 3). At the additional sites examined in this study, this general trend of lower sulfide concentrations in older aggregations appears to hold true. However, this trend is manifest differently in aggregations containing varying proportions of L. luymesi with older L. luymesi dominated aggregations exhibiting lower sulfide concentrations than older aggregations containing more S. jonesi.

L. luymesi and S. jonesi nearly always co-occur, suggesting that they require very similar geochemical conditions for successful recruitment and survival. However, it is possible that the pre-existing conditions at a seepage site determine the relative abundance of L. luymesi and S. jonesi with S. jonesi having the competitive advantage in areas of higher epibenthic sulfide. If environmental characteristics alone determined the relative abundance of tubeworms, one would expect to find reduced S. jonesi abundance in older aggregations with lower epibenthic sulfide concentration. This hypothesis is not

137 supported in the data as the proportion of S. jonesi does not significantly correlate with age (r = 0.103, p = 0.685).

Alternatively, the relative abundance of the two species may be a result of the timing of settlement, rather than the preexisting geochemical conditions of a site. This follows the lottery hypothesis (Sale 1977) where the species with competent larvae present at the time that suitable substrate becomes available will come to dominate the community. If L. luymesi is able to colonize a local seepage source prior to S. jonesi, it may begin the process of root growth and depression of surface expression of sulfide.

This would inhibit the colonization and growth of S. jonesi, which may be more reliant on epibenthic expression of sulfide in its environment. If S. jonesi is the initial colonist, it can dominate the space available prior to extensive root growth by L. luymesi, maintaining elevated concentrations of sulfide in its environment. This ensures the persistence of endemic species in lower trophic levels, as well as its own survival. The common occurrence of both species in tubeworm aggregations throughout the ULS suggests that neither species may completely exclude the other, regardless of the mechanism involved.

It has been suggested that L. luymesi and S. jonesi have different strategies for sulfide acquisition. L. luymesi has been shown to utilize a posterior extension of its body, the “root”, for sulfide uptake (Julian et al. 1999, Freytag et al. 2001). S. jonesi possesses this structure and is likely capable of utilizing it for sulfide uptake, but the relative abundance and properties of its vascular and coelomic sulfide-binding hemoglobins suggest that this species may be better adapted to take up sulfide at the lower environmental concentrations around their plume (Freytag 2003). In addition to uptake

138 through its roots, L. luymesi may further impact its environment by releasing the sulfate generated by its sulfide-oxidizing symbionts into the sediments (Chapter 3). This would supply an energetically favorable oxidant into deeper sediment layers, thereby augmenting integrated rates of anaerobic hydrocarbon oxidation. This would shift the site of sulfide generation deeper within the sediments, potentially reducing the amount of sulfide produced near the sediment surface. Increased rates of hydrocarbon oxidation would also enhance carbonate deposition, further restricting seepage pathways. Taken together, these mechanisms could produce the observed reductions in sulfide concentration in L. luymesi dominated aggregations.

The effects of the interaction of aggregation age and proportion of L. luymesi are evident in the high degree of similarity in the community composition of three aggregations: 1d, 2b, and 7b. These aggregations were collected at three different sites and were widely disparate in age but exhibited consistencies in sulfide concentration.

Sulfide was present at plume level in all 3 aggregations. While aggregation 2b had lower sulfide concentrations than the other 2, even very low concentrations of sulfide (0.16 µM average at plume level) can have adverse effects on species with low sulfide tolerance if sulfide is persistent (Greishaber and Volkel 1998). The youngest of the three aggregations (1d) was composed of 93% L. luymesi, 2b was composed of 69.1% L. luymesi, and aggregation 7b had only 5.4% L. luymesi. It is the interplay of age and relative abundance of L. luymesi which combine to produce similar environmental regimes. Comparable sulfide concentrations give rise to comparable faunas in three tubeworm aggregations with few other conspicuous similarities.

139 While the strongest correlations to community similarity are found with aggregation age, it is sulfide concentration which is the likely proximal cause of these temporal changes. Endemic species are likely to have higher tolerances to environmental sulfide. This tolerance allows them to take advantage of localized chemosynthetic productivity in the absence of high predation pressure (Carney 1994). As sulfide levels subside, species in higher trophic levels can colonize the aggregations and prey upon these small gastropods and crustaceans.

If capable of modifying the abiotic conditions within the physical structure it creates, L. luymesi would act not only as a foundation species providing habitat, but as an autogenic ecosystem engineer. The uptake of sulfide and the release of sulfate into deeper sediment layers, as suggested by previous modeling studies (Chapter 2, 3), allow L. luymesi to control the factor with potentially the most profound effect on the communities at hydrocarbon seeps. It acts to facilitate the colonization of species with lower sulfide tolerance, increasing the rate of succession in these communities.

Aggregations composed of a majority of S. jonesi will reach the later successional stages, but the process is likely to be far slower. The ability of L. luymesi to reduce epibenthic sulfide levels makes it the most significant engineer known from chemosynthetic environments. Further, the effect size of L. luymesi, covering spatial scales from the aggregation to large uninterrupted fields of tubeworms and temporal scales measured in the centuries approaches that of some of the best known examples of ecosystem engineering (Jones et al. 1994). Tilefish burrows are meters wide and persist for years to decades (Coleman and Williams 2002). The discarded shells of mollusk are relatively small, but can persist for centuries (Gutierrez et al. 2003). The engineering behaviors of

140 beavers and prairie dogs may impact areas measured in kilometers and last for centuries

(Naiman et al. 1986, Whicker and Detling 1988). The temporal and spatial scales on which it acts place the engineering capacity of this vestimentiferan tubeworm on similar scales to some of the most prominent examples known.

In summary, sulfide concentration among the tubes of vestimentiferans is related to the relative age of a tubeworm aggregation and the proportion of L. luymesi in the aggregation. These changes in sulfide concentration have a significant impact on the composition and structure of the communities which inhabit them. By altering sulfide concentration within the habitat it creates, L, luymesi acts as both an autogenic and allogenic engineer. The profound influence of L. luymesi on the surrounding community structure demonstrates the significant ecological role of this engineering species in the

Gulf of Mexico.

Notes

Ben Predmore, Chris Jones, and Peter Deines are responsible for the sulfide concentration data included in this study and will be included as authors. Chuck Fisher collected the majority of the samples and assisted in the interpretation of the data and composition of the manuscript. Ben Predmore, Meredith Redding, Julie Barsic, Brian Tiegs, and Nicole

Iacchei all contributed to the enumeration and measurement of tubeworms and associated fauna. I would also like to acknowledge the assistance of Stephane Hourdez, Breea

Govenar, Sharmishtha Dattagupta, Guy Telesnicki, Mike McGinley, Liz Podowski, and the Craig Young lab for their assistance with collections at sea.

Chapter 6

Community Structure of Gulf of Mexico Cold Seeps: Do General Theories of Deep- sea Ecology Pertain to Chemosynthetic Ecosystems?

Abstract

The majority of deep-sea ecological theory is derived from studies of the benthos inhabiting the relatively monotonous abyssal plain and continental slope sediments.

Subsequent exploration of additional habitats, including hydrothermal vents and seamounts, has allowed a more careful examination and revision of the prevailing theories of deep-sea ecology. Here we examine 40 cold seep macrofaunal community collections from 520 to 3300 m depth in the Gulf of Mexico in light of some of the existing theories of deep-sea ecology. Within the upper slope communities, community composition varies with relative age of the aggregation, with similar communities across a 500 km area of the northern Gulf of Mexico. There was little overlap between the upper slope communities and those of the lower slope and shelf base. This suggests a faunal break in these communities somewhere between 900 and 1800 m. Diversity of the associated communities decreased with depth as endemicity increased. Abundance and biomass remained relatively constant throughout the depth range sampled, contrary to findings from soft-sediment communities. Body size of the individuals in commonly occurring families exhibited varied responses to depth with some families increasing and some decreasing in size at deeper seeps. Overall, the ecology of the cold seep communities of the Gulf of Mexico shared more similarities to hydrothermal vents and seamounts than the soft sediment communities of the deep-sea.

142 Introduction

Hypothesis-driven research in the deep-sea began in the late 1960s with the emergence of the stability-diversity hypothesis to explain the unexpectedly high diversity in deep-sea soft-sediment habitats (Sanders 1968). High diversity in the sediments was explained by the stability of the deep-sea environment allowing for fine-scale niche diversification and the avoidance of competition in seemly similar co-existing fauna.

Further studies have shown that there is a diversity peak at moderate water depths at the base of the continental rise, with subsequent declines in diversity as water depth and distance from the productive waters of the continental shelf increases (Pequenat et al.

1990, Cosson et al. 1997). Initial increases in diversity with depth were accompanied by reductions in abundance and body size of organisms (Haedrich and Rowe 1977, Jumars and Gallagher 1982, Lampitt et al., 1986). These were hypothesized to reflect the quantity and quality of the food supply derived from the photic zone of the oceans that reached the ocean floor.

As more detailed information on the distribution of deep-sea species was obtained, consistent breaks in community structure with high levels of species replacement over relatively short depth intervals were found (George and Menzies 1972,

Pequegnat et al., 1990). The rapid turnover of species was hypothesized to be related to the thermal (Pequegnat et al., 1990) or hypoxic (Lampitt et al., 1986, Wishner et al.,

1995) tolerance of the organisms, or barriers to dispersal imposed by boundaries in current flow (George and Menzies 1972, Soto 1991).

143 While these general trends of deep-sea ecology appear to be consistent throughout the benthos of sedimented habitats on the continental slope and abyssal plain (Gage and

Tyler 1991), this background is punctuated by seamounts and hydrothermal vents. These habitats harbor abundant communities exhibiting high degrees of endemicity and relatively low diversity (Tunicliffe 1991). High macrofaunal abundance is due to the elevated food inputs at hydrothermal vents (Sarazin and Juniper 1999) and seamounts

(deForges et al. 2000) resulting from local chemosynthetic primary production or enhanced re-suspension of material by accelerated currents, respectively. Increased energetic input also leads to larger biomass per individual with the “giant” chemosynthetic species (vestimentiferans, mytilid mussels, and vesicomyid clams) dominating hydrothermal vents, and large scleractinian and gorgonian coral colonies occupying seamount habitats. Though the ecology of these habitats appears to be influenced by very different factors, the bathymetric zonation of benthic communities remains consistent. Hydrothermal vent species are found along oceanic ridges over thousands of kilometers, and yet species overlap between sites at different depths along the ridge can be low (Desbruyères et al 2000). There appears to be a consistent

“intertidal-like” zonation of seamounts along sub-surface mountain chains in the South

Pacific (Wishner et al. 1995). This principle of depth zonation appears to be consistent among all deep water habitats.

Ecological studies of cold seeps have mainly been restricted to patterns in community structure over relatively small spatial or bathymetric scales (Juniper and

Sibuet 1987, Olu et al. 1997, Levin et al. 2000, Sahling et al. 2002, Bergquist et al.

2003a, Turnipseed et al. 2003), the exception being the review of cold seep ecology by

144 Sibuet and Olu (1998). Existing data indicates that diversity, density, and rate of endemicity of these communities rival that of hydrothermal vents and seamounts (Sahling et al. 2002). Studies of the communities associated with mussel beds at seeps show that species richness may in fact be much higher than vents and seamounts (Turnipseed et al.

2003). Rates of endemism may also be higher, potentially due to the increased isolation of cold seep sites that lack the geological continuity provided by mid-ocean ridge or seamount chain systems (DeForges et al. 2000, Turnipseed et al. 2003). While cold seeps may be similar to hydrothermal vents in that they are influenced by potentially toxic levels of hydrogen sulfide (Sahling et al. 2002), they are more stable with local seepage sources persisting on the order of centuries rather than years to decades (Carney 1994).

The continental slope of the northern Gulf of Mexico lies in a salt-dome province.

The mobility of the underlying salt sheet causes the vertical migration of hydrocarbons resulting in extensive areas of seepage along the slope (Kennicutt et al. 1988, MacDonald

1998). High concentrations of hydrogen sulfide are often associated with these seepage areas resulting primarily from the reduction of seawater sulfate through anaerobic methane oxidation and hydrocarbon degradation (Aharon and Fu 2000). Expression of hydrogen sulfide and methane at the sediment-water interface has resulted in the presence of diverse communities of symbiotic chemosynthetic organisms exploiting this energetic resource (Kennicutt et al. 1985).

The upper Louisiana slope (ULS) cold seeps are dominated in terms of biomass by the vestimentiferans Lamellibrachia luymesi and Seepiophila jonesi along with the mussel Bathymodiolus childressi (MacDonald et al. 1990). Recent studies of the community ecology of seeps at a few sites clustered within a 5 km radius of the ULS

145 have demonstrated a consistent pattern of community succession at these sites (Bergquist et al. 2003a, Chapter 4). Mussels dominate young seeps with high abundances and low diversities of associated fauna comprised of mainly endemic species (Bergquist et al. in press). In later successional stages, tubeworms dominate the community and harbor more diverse communities of a mixture of endemic and non-endemic species. As seepage rates decline in older tubeworm aggregations, the associated community becomes less abundant and diverse. The seeps of the lower slope and base of the continental shelf of the Gulf of Mexico are inhabited by the tubeworms Escarpia laminata and an unidentified lamellibrachid and 3 species of modiolid mussels, B. childressi, B. heckeri, and B. brooksi (Paull et al. 1984, Bryant et al., 1990, MacDonald et al. 2003). Though there is a paucity of data on communities associated with these chemosynthetic species, the mussel beds of the Florida Escarpment seeps show some of the most highly diverse seep communities yet known (Turnipseed et al. 2003).

In this study, we examine some of the long-standing theories of deep-sea ecology in the cold seep ecosystems of the Gulf of Mexico. We compare the similarity in structure of communities associated with 32 tubeworm aggregations collected at similar depths across 500 km of the upper Louisiana continental slope. We investigate bathymetric patterns in diversity, endemicity, abundance, and biomass of 40 seep-associated communities from 550 to 3300 m depth. We also examine the change in body size of common families with depth. These findings are discussed with respect to our current understanding of deep-sea ecology in general, and hydrothermal vent and seamount communities in particular.

146 Materials and Methods

A total of 28 vestimentiferan aggregations and 8 mussel bed samples were collected on 5 cruises aboard the R/V Seward Johnson II and the R/V Atlantis from 2002 to 2004. Collections were made at 12 different sites on the upper continental slope and 3 different sites on the lower slope of the Gulf of Mexico (Fig. 6-1). The upper slope sites sampled with the Bushmaster collection device in this study were divided into 3 regions: central (5 sites in Green Canyon), eastern (2 sites in Mississippi Canyon and Vioska

Knoll), and western (3 sites in Garden Banks). The central sites have been the subject of numerous studies on the biogeochemistry and community ecology of ULS seeps over the past 20 years (Kennicutt et al. 1985, Carney 1994, Bergquist et al. 2003, Arvidson et al.

2004, Chapter 4, 5). Two of the central sites lie along the same fault scarp (Sager et al.

2003) in the Green Canyon (GC) 234 (site 1; 27:44.8N, 91:13.3W, 530 m depth) and

GC232 (site 2; 27:44.5N, 91:19.1W, 570 m) lease blocks of the ULS. The sites were approximately 10 km apart and ranged from 528 to 571 m depth. They are both dominated by extensive aggregations of Lamellibrachia luymesi and Seepiophila jonesi with the more rare presence of the undescribed escarpid. Interspersed among the tubeworms are beds of the mussel Bathymodiolus childressi. Just north of these sites, in

GC233 (site 3; 27:43.4N, 91:16.8W, 650 m), is the third central site. This site is characterized by a pool of hypersaline brine. The borders of this brine pool are entirely colonized by B. childressi with relatively small aggregations of tubeworms located nearby. The last central site was in GC354 (site 4, 27:35.9N, 91:49.4W, 560 m), approximately 60 km to the west of these sites on the flanks of a large mound. Two

147 samples from the “Bush Hill” site located in GC184 (BH; 27:46.9N, 91:30.5W, 535m) and two samples from GC234 (GC) obtained in 1998 (Bergquist et al. 2003) are included here for comparison.

Figure 6-1: Map of collection sites of communities associated with tubeworm aggregations or mussel beds. Circles surround clusters of sites within the same area of the Gulf of Mexico. Depth contours are in meters.

The eastern sites were in the Mississippi Canyon and Vioska Knoll lease areas designated by the U.S. Minerals Management Service. Site 5 is in the Mississippi Canyon

(MC) 885 lease block approximately 150 km to the east of GC234 (28:03.7N, 89:42.6W,

620 m). This site was characterized by a series of ridges associated with shallowly buried carbonate outcrops with isolated tubeworms and small tubeworm aggregations (mainly L. luymesi). Site 6 lies 200 km to the north-east of MC in the Viosca Knoll (VK) 826 lease block (29:09.3N, 88:01.4W, 540 m) on the flanks of a large mound with abundant seepage and large carbonate blocks. The exposed portions of the carbonate blocks may be colonized by gorgonian (mainly Callogorgia spp.) and scleractinian (mainly Lophelia pertusa) corals or tubeworm (L. luymesi) aggregations. The western sites (sites 7a-d)

148 were a cluster of 3 similar sites in the Garden Banks (GB) lease area located approximately 200 km to the west of GC234 (27:25.9N, 93:08.8W to 27:27.3N,

93:35.3W, 550 to 630 m). This area of Garden Banks is the site of multiple areas of localized brine seepage with high concentrations of lighter hydrocarbons, primarily methane. The tubeworm aggregations at this site were associated with smaller carbonate outcrops protruding above the sediment. Mussel communities were sampled at a heavily brined site in GC204 (site 8; 27:45.1N, 90:33.0W, 835 m) and a site of extensive fluid flux associated with barite chimneys in MC929 (site 9; 28:01.5N, 89:43.6W, 636 m).

The deeper sites of the lower slope are the Florida Escarpment, Atwater Valley, and Alaminos Canyon. The Florida escarpment is a massive carbonate wall on the continental slope off the western edge of Florida at approximately 3300m. Where the base of this wall meets the abyssal plain, extensive areas of hypersaline brine seepage containing high concentrations of hydrogen sulfide are present (Paull et al., 1984). Both vestimentiferan (Escarpia laminata) and mussel (Bathymodiolus heckeri and B. brooksi) communities were sampled at this site (26:01.8N, 84:54.7W). Atwater Valley is a continuation of the Mississippi Canyon area of the upper slope extending from 1800 m to the abyssal plain. The site at AT425 is a large knoll with extensive areas of seepage at the crests of series of ridges along the flanks of a large knoll (MacDonald et al. 2003). The community associated with a mussel bed consisting of Bathymodiolus sp., tentatively assigned to B. brooksi (S. Carney, pers. comm.),was sampled at this site (27:34.1N,

88:29.8W, 1893 m). In addition to mussel beds, a few E. laminata aggregations were noted and an undescribed species of lamellibrachid tubeworm was collected during the same submersible dive. Alaminos Canyon cuts through the continental shelf off of Texas

149 between 2200 and 3000m exposing alternating salt and carbonate layers (Bryant et al.,

1990). Both mussel (B. childressi) and tubeworm (E. laminata) communities were sampled at this site (26:21.3N, 94:29.9W, 2215 m).

Intact vestimentiferan aggregations were obtained using the Johnson Sea-Link I and II and Alvin submersibles with the Bushmaster Jr. and Bushmaster Sr. collection devices (Govenar et al 2002, Bergquist et al. 2003a, Chapter 4, 5). The contents of each

Bushmaster collection were sieved through a 2 mm mesh and sorted to the lowest possible taxonomic level on board the ship. Mussel bed samples were obtained with a

“clam shell” sampler manipulated by the submersible. This sampling device consists of an aluminum frame with rounded plexiglass sides that rotate to enclose the sample. Each mussel bed was sampled with 3-4 successive grabs of 342 cm2 surface area from this sampling device. Samples were sieved through a 2 mm mesh and processed similarly to tubeworm samples. Mussel community data were included in the study as relative abundance due to the imprecise quantification of sampling volume.

All associated fauna were fixed in formalin and preserved in ethanol for transport back to Penn State and final determination of taxonomic status. E.E.C. and S.H. identified all polychaetes. Primary identification of other groups was carried out and specimens sent to experts for further identification or verification. All individuals were weighed and enumerated for subsequent analyses. Preserved wet weight was converted to ash free dry weight (AFDW) using existing species specific conversion factors (Bergquist et al. 2003a) when available, or by published higher taxonomic level conversion factors

(Ricciardi and Bourget 1998). Species are reported as endemic if they have only been collected at cold seeps or contain chemosynthetic or methanotrophic symbionts.

150 Shannon-Weaver diversity index ( H '= − p ln p ) and Pielou’s index of evenness ∑i i i

( J '= H '/ ln S ) were calculated for each collection. Colonial organisms contributed to species richness, but were not enumerated and therefore were not included in quantitative analyses.

Tubeworms were counted and measured on board, time permitting. The remainder of the aggregations were preserved in formalin and transported back to Penn State for processing. Length was measured to a standardized posterior outer tube diameter, 2 mm for L. luymesi and E. laminata and 4 mm for S. jonesi. This diameter was the common point where the tubeworms entered a dense, tangled mass at the base of the aggregations.

Efforts to measure the species to a smaller diameter resulted in a significant loss of quantitative data on tube length due to tube breakage. Surface area was calculated as for a cone frustrum:

1 SA = π ⋅l()AD 2 + PD 2 + AD ⋅ PD (1) 3 where SA = surface area, l = length, AD = anterior diameter, and PD = posterior diameter.

Published conversions of length to AFDW were used to calculate biomass of tubeworm species (Bergquist et al. 2003a).

Similarity in community structure between aggregations was examined using multidimensional scaling (MDS) using PRIMER software (PRIMER-E ltd.). This analysis uses an iterative procedure to minimize the error in a monotonic regression between ranked Bray-Curtis (BC) similarity values and 2-dimensional distance between each pair of samples. Bray-Curtis similarity was determined from the following function:

151

⎛ p ⎞ ⎜ ∑ yij − yik ⎟ Sjk = 100⎜1− i=1 ⎟ (2) ⎜ p ⎟ ⎜ ∑()yij + yik ⎟ ⎝ i=1 ⎠ where yij is the abundance of the ith species in the jth sample and p is the total number of species. For Bray-Curtis analysis of upper slope tubeworm samples, species abundance was standardized to tubeworm surface area to provide a measure of species density irrespective of collection size. Density data were 4th root transformed to reduce the bias towards the most dominant species (Clarke and Warwick 2001). For Gulf-wide comparisons of community structure, family-level abundance data were used as there was very little species overlap between the upper and lower slope communities. Relative abundance (proportion of individuals in each family) was used to compare community structure to avoid the confounding factors of size of collection and form of biogenic habitat (tubeworm vs. mussel).

To assess the relative influence of abiotic factors on community similarity, a modified BIO-ENV procedure (Clarke and Warwick 2001) was used. Similarity

(Euclidean distance) in depth and distance between collections was determined.

Spearman rank correlation between Bray-Curtis similarity in community structure and

Euclidean distance in habitat characteristics was used to determine the environmental factor that best explained the pattern in community similarity.

152 Results

There were a total of 162 species collected in the 40 samples taken, including 11 symbiotic species of vestimentiferans and bivalves (Appendix). Of these 18 are confirmed to be new species awaiting description. In addition, a new species of lamellibrachid tubeworm was collected at Atwater Valley. This species had been previously observed (MacDonald et al. 2002), and may be the same species of undescribed lamellibrachid previously collected from the Alaminos Canyon (Nelson and

Fisher 2000), though this determination awaits further taxonomic investigation. The 32 upper slope samples contained 116 species, while the 8 lower slope samples contained 48 species. There was little overlap in the bathymetric distributions of the species found on the upper and lower slope. The only species obtained in both sets of collections were the polychaetes Eunice sp. nov., Eurythoe sp. nov., and Nereis sp. nov., the sipunculan

Phascolosoma turnerae, and the symbiotic mussel Bathymodiolus childressi. All other species were restricted to either shallower than 840 m or deeper than 1890 m.

Similarity in structure of tubeworm associated communities throughout the ULS is shown in the multi-dimensional scaling (MDS) plot ( Fig. 6-2 ). Community similarity was most tightly correlated to similarity in relative aggregation age (r = 0.300, p < 0.001), with relative age generally increasing from left to right along the x-axis. The relative importance of age in explaining community structure is evidenced by the interspersion of collections from site 1 (GC234) throughout the multivariate community space, and the clustering of communities inhabiting similarly aged aggregations from multiple sites (1d,

2b, 2c, 3a, 7a, 7b). There was some measure of geographic substructure within this depth

153 stratum as aggregations located closer together tended to have more similar communities

(r = 0.143, p = 0.004). The most similar communities were 1a and b which were collected mere meters from one another. The next most similar communities were sampled in

“young” tubeworm aggregations (Bergquist et al 2003a, Chapter 4) from two nearby sites

Bush Hill (bh2) and GC234 (1g). The one region of the plot (lower left of Fig. 2) that is not occupied by collections from the central sites (GC234, GC232, Bush Hill), contains similar communities from sites located both to the west and east of the central sites.

These include similar communities from a western site in Garden Banks (7d) and a site located on the western edge of Green Canyon, GC354 (4a), and two communities from

Vioska Knoll (6a and b) located well to the east.

154

Figure 6-2: Multidimensional scaling (MDS) plot of similarity among associated community samples from the upper Louisiana slope. Collections are arranged according to their rank similarity to each of the other samples with closer symbols with higher community similarity. Stress value (0.17) indicates a general agreement between distance on the ordination and difference in community structure. Circles indicate groups of samples with greater than 55% Bray-Curtis community similarity and are included as an assessment of the validity of the ordination.

Along the upper slope, the majority of species were present in all three regions of the slope. There were few species whose distributions were limited to east or west of the central sites (Appendix). The distributions of Acesta bulli, Callogorgia americana,

Cantrainea macleani, Rochina crassa, Sabelliastarte sp. and Tamu fisheri were restricted to the central and eastern sites, while Bellotia sp., Ectomyxilla methanophila, Eumida sanguinea, Eunice sp nov., Euthalanessa sp., Lumbrineris tenuis, Neoamphitrite sp., and

Phippsia sp. 2 were restricted to the central sites and those to the west. These represent

155 only 11% of the species collected from the upper slope, and were not among the 10 most common species in any of these collections. Seep-endemic species comprised a relatively high proportion of all communities. Endemicity ranged from 23.8 to 61.5% for the upper slope and 40 to 100% for the deeper sites.

The MDS plot of all community collections shows the distinct separation of the communities into the broad habitat types sampled (Fig. 6-3). At the family level, the upper slope tubeworm communities are arranged in a very similar configuration to the species level MDS plot (Fig. 6-2) with the youngest aggregations to the left side of the multivariate space and older aggregations to the right. Upper slope mussel communities are similar to each other, occupying a restricted portion of the 2-dimensional community space. However, upper slope (524 to 835 m depth) mussel and young tubeworm communities share a number of common species including B. naticoidea, C. macleani, C. meroglypta, A. stactophila, and P. turnerae.

All of the communities from the lower slope cluster together, with mussel beds and tubeworm communities overlapping. While the qualitative family-level composition of the upper and lower slope communities is similar, the shared families comprise different proportions of the communities. Endemic gastropods (neritoidea) and galatheid crabs often dominate the upper slope communities, while trochid gastropods and phyllodocids often dominate the lower slope habitats. Bresilid shrimp (Alvinocaris spp.) were abundant members of all habitats. Lower slope communities were often numerically dominated by the ophiuroid Ophiurocten spinilimbatum, while the upper slope contains two rare species of brittle star, a juvenile specimen of Amphioplus sp. and an undescribed ophiuroid normally associated with the gorgonian Callogorgia americana.

156

Figure 6-3: Multidimensional scaling (MDS) plot of similarity among associated community samples from all regions of the Gulf of Mexico. Collections are arranged according to their rank similarity to each of the other samples. The stress value of 0.2 indicates a reasonable fit between the ordination and similarity matrix. It should be noted that the upper slope collections are shown in the lower right region of the figure in a similar orientation to Fig. 6-2.

There was a general decline in diversity of cold seep associated communities with depth. The range of diversity for the upper slope communities (H’ = 1.369 to 3.118) was higher than for the lower slope (H’ = 0.022 to 2.029) and the base of the continental shelf

(H’ = 1.024 to 1.861) (Fig. 6-4). Diversity on the upper slope was significantly greater than in the deeper communities (2 sample t-test; T = 4.23, p = 0.0029). Diversity of

157 mussel associated communities consistently fell in the lower end of the range of diversities found at any sampling depth. Community evenness followed a similar trend with higher estimate for the upper slope (J’ = 0.423 to 0.932) than the deeper sites (J’ =

0.032 to 0.689), with this difference also statistically significant (T = 3.68, p = 0.0063).

0

-500

-1000 ) -1500

depth (m -2000

-2500

-3000

-3500 0123400.51 diversity (H') evenness (J')

Figure 6-4: Diversity of communities associated with tubeworm aggregations (·) and mussel beds (*) throughout the Gulf of Mexico. a. Change in diversity (H’) with depth. b. Change in evenness (J’) with depth.

Abundance and biomass of macrofauna inhabiting cold seeps does not appear to decline with depth. Density ranges from 11 to over 5600 individuals per m2 tube or shell surface area (Fig. 6-5a). The cold seep communities at the deeper sites fall in the middle of this range from 134 to 4457 ind·m-2. Biomass of tubeworm associated communities does appear to decline with depth with the deeper water communities supporting biomasses at the lower end of the range of the upper slope communities (Fig. 6-5b). The

158 mussel communities had higher biomass, approaching that of the upper end of the range of upper slope tubeworm communities.

0

-500

-1000

) -1500

depth (m -2000

-2500

-3000

-3500 01234-10123

2 2 density (ind/m ) biomass (gAFDW/m )

Figure 6-5: Size of communities associated with tubeworm aggregations (·) and mussel beds (*). a. Density of associated fauna shown on log scale. b. Biomass of associated fauna shown on log scale.

Average body size showed varied responses to changes in depth among the families for which this could be assessed (Fig. 6-6). Of the 8 families that occurred frequently throughout the depth range sampled, the bresilidae, galatheidae, and to a lesser extent the phyllodocidae showed larger body sizes at the deeper sites. The polynoid polychaetes and trochid gastropods were the only families to follow the predicted trend of smaller body sizes with increasing depth.

159

bresilid galtheid phyllodocid polynoid trochid -3 -2 -1 0 -3 -2 -1 0 -5 -4 -3 -2 -1 0 -4 -3 -2 -1 0 -4 -3 -2 -1 0 0

-500

-1000

-1500

-2000 depth (m)

-2500

-3000

-3500

Figure 6-6: Change in body size with depth of 5 common families of seep inhabitants. Body size is expressed as biomass (AFDW) per individual shown on log scale.

Discussion

In this study, long standing theories of deep-sea ecology were examined in 40

cold seep community collections from the Gulf of Mexico. The results of this study were

not consistent with many of these theories, similar to the manner in which hydrothermal

vent and seamount communities stand in contrast to the ecology of the surrounding

benthos. Diversity at the seeps examined was relatively low as compared to the general

benthos, and approximates that found at vents and seamounts. It should be noted that the

macrofaunal species richness of the seep communities reported here (162 species retained

on a 2 mm sieve) rivals that of all known cold seep associated species at the time of the

last comprehensive review (211, Sibuet and Olu 1998). Diversity tended to decline with

160 depth, while rate of endemicity increased. Overall density and biomass of fauna did not show a significant change with depth. Similarly, average body size within families showed a variety of trends, depending on the group examined. There was a significant faunal break detected in this study, with the communities shallower than 800 m having few species in common with those below 1800 m. While a few aspects of the ecology of

Gulf of Mexico seeps fit the model of the general ecology of the deep-sea, the majority of the community characteristics did not.

The distribution of upper slope communities within the MDS plot is determined mainly by the age of the tubeworm aggregations. Results from previous studies of collections from sites 1 and 2 (GC234 and GC232) showed that age is the most significant factor in determining community structure (Chapter 4). This follows a predictable successional sequence of community replacement (Bergquist et al. 2003a), a model that was originally based on the 4 collections from GC234 (gc3, gc4) and Bush

Hill (bh2, bh3) included here for comparison. The interspersion of collections from other sites within the distribution of central site communities in the multivariate community space suggests that the same successional processes guide community development at cold seeps across the northern Gulf of Mexico.

The similarity in the communities distributed across 500 km of the ULS is mainly due to consistent presence of common seep-endemic species throughout the sampling area. High rates of endemicity arise from the exclusionary chemical environment encountered of the seeps. High concentrations of sulfide, methane, oil, and polyaromatic hydrocarbons (Sassen et al. 1994, Arvidson et al. 2004) in conjunction with anoxic or hypoxic conditions (Hourdez et al. 2002) select for species well adapted to the conditions

161 encountered at seeps. The influence of sulfide on the community structure is more prevalent in younger aggregations where sulfide is more abundant (Bergquist et al.

2003b, Chapter 5). This leads to a consistent suite of endemic species in early successional stages, which is reflected in the high degree of similarity in young communities (upper left of Fig. 2). In older aggregations, sulfide concentrations decline

(Bergquist et al 2003b, Chapter 5 and the “colonist” and “vagrant” fauna of the Gulf

(Carney 1994) are able to colonize the seeps (Bergquist et al. 2003a, Chapter 4). This leads to a lower community similarity values among older aggregations (1j, 1f, gc4) as colonization by background species is a more random process.

The variability in the background fauna results in a certain degree of affinity within site, particularly in the most isolated sites, Vioska Knoll (site 6) and to a lesser extent, Garden Banks (site 7). These smaller scale differences in community structure are likely due to local environmental characteristics such as seepage rate, topography, and current structure selecting for species with comparable habitat preferences. The Garden

Banks (site 7) and GC233 (site 3) sites have the common feature of containing methane rich pore fluids supporting populations of the mussel with methanotrophic symbionts,

Bathymodiolus childressi, within their communities. Site 3 is the location of a large dense pool of hypersaline brine on the seafloor supporting vast areas of B. childressi mussel beds. Vioska Knoll (site 6) and GC354 (site 4) share the commonality of supporting communities of the scleractinian coral, Lophelia pertusa. While there is very little direct overlap in the communities harbored by the corals and the tubeworms (E.E.C. unpublished data), the similarity within the tubeworm communities suggests that the

162 large physical structure provided by the corals may exert some influence on the community structure of these sites.

The consistency of the fauna inhabiting ULS seeps is comparable to the ecology of hydrothermal vents and seamounts where the communities exhibit high similarity across thousands of kilometers along oceanic ridges or seamount chains (deForges et al.

2000). The similarity in the fauna inhabiting vent systems is a result of relatively high rates of endemicity, with some species inhabiting the hydrothermal vents of the East

Pacific Rise also present at the Juan de Fuca sites, nearly 5000 km distant (Tsurumi and

Tunicliffe 2001). While the rate of endemicity at ULS seeps is on the same order of that of seamounts (29-34%, de Forges et al. 2000) and vents (up to 95%, Tunicliffe 1991), it is likely to be even higher than reported here since the taxonomic status, and therefore the habitat specificity, of many of the species in the study remains unresolved. These deep- sea ecosystems exhibit analogous patterns in community structure though the mechanism generating the pattern is different. Deep-sea chemosynthetic systems are isolated by their high concentrations of reduced chemical compounds (Johnson et al. 1988, Turnipseed et al. 2003), while seamounts are isolated by physical barriers to dispersal and migration (de

Forges et al. 2000). Current dynamics associated with seamount chains provide ample dispersal capabilities along the axis of the seamount chain, but very little exchange between parallel chains (de Forges et al. 2000).

The general lack of geographic substructure in the seep communities spanning

500 km across the northern Gulf of Mexico suggests that there is little barrier to dispersal or even migration among these sites. The low number of species with distributions restricted to the east or west of the central sites reflects this trend. Even those few species

163 that appeared to have restricted distributions may inhabit the entire slope, having not been captured in the eastern or western regions due to limited sampling effort in these areas.

Genetic studies of L. luymesi are also consistent with a lack of a physical dispersal barrier. While some population substructure exists, tubeworms throughout the upper

Louisiana slope appear to be freely interbreeding (McMullin 2003). This similarity in genetic structure is mirrored here in the similarity in community structure.

While the communities showed relatively low levels of variability along the upper slope down to the GC204 mussel bed collection at 835 m, there were few similarities between depth ranges. There were some similarities in community composition at the family level, but trophic structure was quite different. Bresilid shrimp (Alvinocaris spp.) were common and abundant members of the seep community throughout the depth range sampled. The most striking difference was the dominance of the first-level consumer trophic position by Ophiurocten spinilimbatum in the deeper communities rather than grazing gastropods or galatheid crabs. The only species that crossed the boundary between the upper and lower slope were 3 undescribed species of polychaetes, the sipunculan Phascolosoma turnerae and the mussel Bathymodiolus childressi. Specimens of the three species of polychaetes and the sipunculan from the different depth intervals were morphological indistinguishable, though their exact taxonomic status awaits additional morphological and genetic investigations by experts in their respective groups.

It is possible that the actually represent different species with similar morphologies, or reproductively isolated populations. In the case of B. childressi, the fairly high degree of genetic divergence between the ULS and AC populations (Craddock et al. 1995),

164 suggests that they may not be freely interbreeding, though they are currently considered to be conspecific (Gustafson et al. 1998).

The depth zonation of the hydrocarbon seep communities mirrors that found in the soft bottom communities of the GoM (Pequenat et al. 1990). In the soft bottom infaunal and epifaunal communities, there was rapid species replacement at 450-500 m,

750-850 m, 1050 m, 2250 m, and 3250m. These breaks roughly coincide with the stratification of the water column with rapid changes in temperature and salinity at 600 m and 1000 to 1200 m. At the seeps, the only faunal breaks that may be confirmed by this study are the 450-500 m break above which cold seeps inhabited by tubeworms and mussels do not exist (approximately 400 m, Carney 1994), and the 1050 m break. There was no evidence for a 2250 m break in the seep communities here, as 13 of the 22 species that were collected more than once occurred at both Florida Escarpment (3300 m) and

Alaminos Canyon (2200 m). The restricted depth range of the seeps coupled with the depth limits of the submersible used for the majority of this study (Johnson Sea-Link, 990 m) precluded investigations of depth zonation on a finer scale.

The similarity between the seep communities of the upper slope and those found at seeps in the Barbados accretionary prism between 1300 and 2100 m (Olu et al. 1997) lends insight into the bathymetric zonation of the seeps of the region. Seep communities sampled at 1300 m near Barbados consisted of Lamellibrachia sp. aggregations along with Bathynerita naticoides, Cataegis meroglypta, Alvinocaris cf. stactophia (Olu et al.

1997). The dominant fauna at this 1300 m site appears similar to that of the upper slope of the Gulf of Mexico from 520 to 835 m and suggests that there is not a pronounced faunal break between these depths. The Barbados sites at 1700 and 2080 m depth were

165 dominated by Escarpia laminata and Bathymodiolus cf. heckeri and contained high abundances of Alvinocaris muricola (Olu et al. 1997), suggesting a greater affinity to the deeper (1890 to 3300 m) sites of the Gulf of Mexico. However, Bathynerita naticoides was found at 1700 m near Barbados, and Phascolosoma cf. turnerae was found at 2080 m, in addition to a number of species of uncertain taxonomy including a species of

Munidopsis, and a number of shared polychaete families that could be common to all depth strata. More intensive quantitative sampling is required within this depth range to resolve the bathymetric distribution of these species.

The commonality of the fauna between the upper slope and Barbados suggests that there is little barrier to dispersal between these two widely separated communities.

The Florida Escarpment site also shares a number of species with the mussel beds found on the Blake Ridge at 2155 m (Van Dover et al. 2003). There were 9 species in common between these two sites, and 5-10 species in common to the Blake Ridge and Barbados sites (Van Dover et al. 2003). This suggests that the current structure of the south Atlantic and Gulf of Mexico provides a mechanism for dispersal among all of these seeps, even if it is a rare event. The loop current of the Gulf is derived from the encroachment of the

Gulf Stream into the Gulf of Mexico though the 1650 m deep Yucatan Strait (Welsh and

Inoue 2000). This would deliver propagules from the Barbados seeps via the Caribbean

Sea to the Gulf of Mexico. The outlet of the gulf is through the Straits of Florida, which are only about 800 m deep (Carney 1994). While this depth is sufficient for dispersal from the ULS sites, the loop current rarely extends far enough west to entrain waters from these sites (Sturges and Evans 1983). The few species shared between Florida

Escarpment and Blake Ridge may have the common feature of producing larvae with

166 fairly high buoyancy and temperature tolerance allowing dispersal across the barrier imposed by the Straits of Florida. The alternative explanation is that the Barbados seeps serve as sources of propagules for both the Gulf seeps as well as the Blake Ridge seeps with little direct interbreeding between the Gulf and Blake Ridge. This remains a viable hypothesis and awaits further investigation.

The decline in seep community diversity below 1800 m follows the trend of diversity for the Gulf of Mexico in general, with the most diverse communities around

1400 m (Pequenat et al. 1990). While the depth of greatest diversity is similar, only a subset of the background fauna has thus far been sampled in the seep habitat. Greater input of allochthonous organic matter to upper slope seeps leads to a variety of nutritional inputs supporting a diverse mixture of endemic species and non-endemic background fauna. The communities inhabiting the deeper sites are almost solely comprised of endemic species because of the lower diversity of background fauna available to colonize these habitats. This phenomenon has also been observed at hydrothermal vent sites at different depths along the mid-Atlantic ridge, with much higher endemicity at deeper sites (Desbruyères et al 2000). However, this change in endemicity was attributed to the increased toxicity of venting fluids under higher pressure (Desbruyères et al 2000), which is not ruled out as a potential explanation for the patterns observed here.

In contrast to diversity, abundance and biomass of seep fauna do not appear to decline with depth. Unlike allochthonous photosynthetic productivity, the quantity and quality of chemosynthetic primary production should not change with depth. This food source supports high numbers of endemic species in lower trophic levels. Endemic species are able to avoid food limitation by exploiting the local resources available, while

167 the background fauna will be excluded from this niche by their lack of tolerance for the abiotic conditions encountered at the seeps. This leads to declines in diversity with depth, but not in the overall abundance or biomass of the communities.

A similar trend is borne out in the body size data of majority of the families in which it could be investigated. Two of the families, the polychaete family polynoidae and the gastropod family trochidae, demonstrated a trend of smaller body size in deeper communities, as would be predicted for the background fauna of the deep-sea. These are scavenging or predatory families that may be responding to the lower overall biomass of potential prey items on the surrounding soft bottom. Some of the families, including the common seep inhabitants of the bresilid shrimp (Alvinocaris spp.) and the galatheid crabs

(Munidopsis spp.), actually showed increases in the average mass of an individual. The lack of a response to depth is a result of their reliance on abundant chemosynthetic production. It is possible that a reduction in competition for autochthonous production allows these families to actually increase in size with depth.

The other family that had larger members in deeper waters was the phyllodocid polychaetes. The increase in per capita biomass seen here is likely due to the alternative life-history that the deeper species exhibits. This species appears to be an ectoparasite on

Escarpia laminata. The digestive cavities of this polychaete were always full of hemoglobin-rich blood, and they were only collected within the tubes of E. laminata. It was inferred from this evidence that they were heamotrophic commensals or parasites on the tubeworms. Additional studies are underway to determine the exact relationship between these polychaetes and their hosts.

168 In this study, we found that some of the prevailing concepts of deep-sea ecology derived from sedimented habitats do not pertain to the cold seep fauna of the Gulf of

Mexico. It appears that these seeps share a similar ecology to hydrothermal vent and seamount habitats rather than the surrounding benthos of the deep-sea sedimented habitats. Diversity generally declined with depth as would be predicted from deep-sea ecological theory, and there were faunal breaks consistent with those found throughout the world’s oceans. However, trends in the abundance and biomass of communities, and body size of many families of endemic fauna did not decline with depth, contrary to other deep-sea habitats. Additional studies of cold seeps along bathymetric gradients in different regions of the world are required to verify the generality of these trends, though the apparent consistency in the ecology of these communities with those of hydrothermal vents and seamounts suggests that they are common phenomena among these specialized habitats.

Notes

Stephane Hourdez, Bob Carney, Roger Sassen, and Chuck Fisher will be coauthors on this study and will assist with preparation of the manuscript for publication. Stephane

Hourdez assisted with species identifications. Bob Carney, Roger Sassen, and Chuck

Fisher led the cruise on the R/V Atlantis when the deep-water data was collected and contributed to the qualitative assessment of the community and habitat characteristics of the sites. Ben Predmore, Meredith Redding, Julie Barsic, Brian Tiegs, and Nicole Iacchei all contributed to the enumeration and measurement of tubeworms and associated fauna.

Bibliography

Aharon P, Fu B (2000) Microbial sulfate reduction rates and oxygen isotope fractionations at oil and gas seeps in deepwater Gulf of Mexico. Geochim Cosmochim Atca 62: 233-246.

Aharon P, Fu B (2003) Sulfur and Oxygen isotopes of coeval sulfate-sulfide in pore fluids of cold seep sediments with sharp redox gradients. Chem Geol 195: 201- 218.

Aitken CM, Jones DM, Larter SR (2004) Anaerobic hydrocarbon biodegradation in deep subsurface oil reservoirs. Nature 431: 291-294.

Aliosi G, Bouloubassi I, Heijs SK, Pancost RD, Pierre C, et al. (2002) CH4-consuming microorganisms and the formation of carbonate crusts at cold seeps. Earth Planet Sci Lett 203: 195-203.

Arp AJ, Childress JJ (1981) Blood function in the hydrothermal vent vestimentiferan tube worm. Science 213: 342-344.

Arp AJ, Childress JJ (1983) Sulfide binding by the blood of the hydrothermal vent tube worm Riftia pachyptila. Science 219: 295-297.

Arp AJ, Childress JJ, Vetter RD (1987) Sulfide binding protein in the blood of Riftia pachyptila is the extracellular hemoglobin. J Exp Biol 128: 139-158.

Arvidson RS, Morse JW, Joye SB (2004) The sulfur biogeochemistry of chemosynthetic cold seep communities, gulf of Mexico, USA. Mar Chem 87: 97-119.

170

Bergquist DC, Williams FM, Fisher CR (2000) Longevity record for deep-sea invertebrate. Nature 403: 499-500.

Bergquist DC, Urcuyo IA, Fisher CR (2002) Establishment and persistence of seep vestimentiferan aggregations from the upper Louisiana slope of the Gulf of Mexico. Mar Ecol Prog Ser 241: 89-98.

Bergquist DC, Ward T, Cordes EE, McNelis T, Howlett S, et al. (2003a) Community structure of vestimentiferan-generated habitat islands from upper Louisiana slope cold seeps. J Exp Mar Biol Ecol 289: 197-222.

Bergquist DC, Andras JP, McNelis T, Howlett S, van Horn MJ, Fisher CR (2003b) Succession in Gulf of Mexico cold seep communities: the importance of spatial variability. PSZN Mar Ecol 24: 31-44.

Bergquist DC, Fleckenstein C, Knisel J, Begley B, MacDonald IR, Fisher CR (in press) Variations in seep mussel bed communities along physical and chemical environmental gradients.

Bertness MD, Leonard GH (1997) The role of positive interactions in communities: lessons from intertidal habitats. Ecology 78:1976-1989.

Boetius A, Ravenschlag K, Schubert CJ, Rickert D, Widdel F, Gieseke A, Amann R, Jorgensen BB, Witte U, Pfannkuche O (2000) A marine microbial consortium apparently mediating anaerobic oxidation of methane. Nature 407: 623-626.

Bosch C, Grasse PP (1984) Cycle partiel des bacte´ ries chimioautotrophes symbiotiques et leurs rapports avec les bacte´ riocytes chez Riftia pachyptila Jones (Pogonophore Vestimentifere). I. Le trophosome et les bacte´ riocytes. C R Acad Sci Paris (Ser III) 299:371–376

171

Bright M, Keckeis H, Fisher CR (2000) An autoradiographic examination of carbon fixation, transfer, and utilization in the Riftia pachyptila symbiosis. Mar Biol136: 621-632.

Brooks JM, Kennicutt MC, Fisher CR, Macko SA, Cole K, Childress JJ, Bidigare RR, Vetter RD (1987) Deep-sea hydrocarbon seep communities: Evidence for energy and nutritional carbon sources. Science 238: 1138-1142.

Brooks JM, Wiesenburg DA, Roberts H, Carney RS, MacDonald IR, Fisher CR, Guinasso NL, Sager WW, McDonald SJ, Burke RA, Aharon P, Bright TJ (1990) Salt, seeps and symbiosis in the Gulf of Mexico. Eos 71: 1772-1773.

Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends Ecol Evol 18: 119-125.

Bryant WR, Bryant JR, Feeley MH, Simmons GR (1990) Physiographic and bathymetric characteristics of the continental slope, northwest Gulf of Mexico. Geo-Mar Lett 10: 182-199.

Carney RS (1994) Consideration of the oasis analogy for chemosynthetic communities at Gulf of Mexico hydrocarbon vents. Geo-Mar Let 14: 149-159.

Carson B, Holmes ML, Umstattd K, Strasser JC, Johnson HP (1991) Fluid expulsion from the Cascadia accretionary prism: evidence from porosity distribution direct measurements, and GLORIA imagery. Phil Trans R Soc Lond Ser A 335: 331- 340.

Cary SC, Fisher CR, Felbeck H (1988) Mussel growth supported by methane as sole carbon and energy source. Science 240: 78-80.

172

Cary C, Fry B, Felbeck H, Vetter RD (1989) Multiple trophic resources for a chemoautotrophic community at a cold water brine seep at the base of the Florida Escarpment. Mar Biol 100: 411-418.

Case TJ (2000) An Illustrated Guide to Theoretical Ecology. Oxford University Press, Oxford. 464 pp.

Castilla JC, Lagos NA, Cerda M (2004) Maine ecosystem engineering by the alien ascidian Pyura praeputialis on a mid-intertidal rocky shore. Mar Ecol Prog Ser 268: 119-130.

Cavanaugh CM, Levering PR, Maki JS, Mitchell R, Lidstrom ME (1987) Symbiosis of methylotrophic bacteria and deep-sea mussels. Nature 325: 346-348.

Childress JJ, Fisher CR, Brooks JM, Kennicutt MC, Bidigare R, Anderson AE (1986) A methanotrophic marine molluscan (Bivalvia: Mytilidae) symbiosis: Mussels fueled by gas. Science 233: 1306-1308.

Childress JJ, Fisher CR (1992) The biology of hydrothermal vent animals: physiology, biochemistry, and autotrophic symbioses. Oceanog Mar Biol Ann Rev 30: 337- 441.

Clarke KR, Warwick RM (2001) Change in Marine Communities: An Approach to Statistical Analysis. Primer-E, Plymouth, UK. 170 pp.

Coleman FC, Williams S (2002) Overexploiting marine ecosystem engineers: potential consequences for biodiversity. Trends in Ecology and Evolution 17(1):40-44.

Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science 199: 1302- 1310

173

Cosson N, Sibuet M, Galeron J (1997) Community structure and spatial heterogeneity of the deep-sea macrofauna at three contrasting stations in the tropical Northeast Atlantic. Deep-sea Res 44: 247-269.

Craddock C, Hoeh WR, Gustafson RG, Lutz RA, Hashimoto J, Vrijenhoek RJ (1995) Evolutionary relationships among deep-sea mytilids (Bivalvia: Mytilidae) from hydrothermal vents and cold-water methane/sulfide seeps. Mar Biol 121: 477-485.

Dando PR, Ridgway SA, Spiro B (1994) Sulphide 'mining' by lucinid bivalve molluscs: Demonstrated by stable sulphur isotope measurements and experimental models. Mar Ecol Prog Ser 107: 169-175. deForges BR, Koslow JA, Poore GCB (2000) Diversity and endemism of the benthic seamount fauna in the southwest Pacific. Nature 405: 944-947.

Delille D, Dellile B, Pelletier E. (2002) Effectiveness of bioremediation of crude oil contaminated subantarctic intertidal sediment: the microbial response. Microb Ecol 44: 118-126.

Desbruyeres D, Toulmond A (1998) A new species of hesionid worm, Hesiocaeca methanicola sp. nov. (Polychaeta: Hesionidae), living in ice-like methane hydrates in the deep Gulf of Mexico. Cah Biol Mar 39: 93-98.

Desbruyeres D, Almeida A, Biscoito M, Comtet T, Khripounoff A, Le Bris N, Sarradin PM, Segonzac M (2000) A review of the distribution of hydrothermal vent communities along the northern Mid-Atlantic Ridge: dispersal vs. environmental controls. Hydrobiologia 440: 201-216.

Distel DL, Baco AR, Chuang E, Morrill W, Cavanaugh C, Smith CR (2000) Do mussels take wooden steps to deep-sea vents? Nature 403: 725-726.

174

Distel DL, Felbeck H (1987) Endosymbiosis in the lucinid clams Lucinoma aequizonata, Lucinoma annulata and Lucina floridana : A reexamination of the functional morphology of the gills as bacteria-bearing organs. Mar Biol 96: 79-86.

Dufour SC, Felbeck H (2003) Sulphide mining by the superextensile foot of symbiotic thyasirid bivalves. Nature 426: 65-67.

Fahse L, Wissel C, Grimm V. (1998) Reconciling classical and individual-based approaches in theoretical population ecology: a protocol for extracting population parameters from individual-based models. Am Nat 152: 838-852.

Felbeck H, Jarchow J (1998) The influence of different incubation media on the carbon transfer from the bacterial symbionts to the hydrothermal vent tube-worm Riftia pachyptila. Cah Biol Mar 39: 279-282.

Ferrell RE, Aharon P. (1994) Mineral assemblages occurring around hydrocarbon vents in the northern Gulf of Mexico. Geo-Mar Let 14: 74-80.

Fisher CR (1990) Chemoautotrophic and methanotrophic symbioses in marine invertebrates. Rev Aquat Sci 2: 399-436.

Fisher CR (1993) Oxidation of methane by deep-sea mytilids in the Gulf of Mexico. In: Biogeochemistry of global change: Radiatively active trace gasses. ed: Oreland RS, Chapman and Hall, New York. pp. 606-618.

Fisher CR (1996) Ecophysiology of primary production at deep-sea vents and seeps. Biosys Ecol Ser 11: 313-336.

Fisher CR, Childress JJ, Oremland RS, Bidigare RR (1987) The importance of methane and thiosulfate in the metabolism of the bacterial symbionts of two deep-sea mussels. Mar Biol 96: 59-71

175

Fisher CR, Kennicutt MC II, Brooks JM (1990) Stable carbon isotopic evidence for carbon limitation in hydrothermal vent vestimentiferans. Science 247: 1094-1096.

Fisher CR, Childress JJ (1992) Organic carbon transfer from methanotrophic symbionts to the host hydrocarbon-seep mussel. Symbiosis 12: 221-235.

Fisher CR, Brooks JM, Vodenichar JS, Zande JM, Childress JJ, Burke RA Jr. (1993) The co-occurrence of methanotrophic and chemoautotrophic sulfur-oxidizing bacterial symbionts in a deep-sea mussel. Mar Ecol 14: 277-289.

Fisher CR, Urcuyo IA, Simkins MA, Nix E. (1997) Life in the slow lane: growth and longevity of cold-seep vestimentiferans. Mar Ecol 18: 83-94.

Fisher CR, MacDonald IR, Sassen R, Young CM, Macko SA, Hourdez S, Carney RS, Joye S, McMullin E. (2000) Methane ice worms: Hesiocaeca methanicola colonizing fossil fuel reserves. Naturwissenschaften 87: 184-187.

Formolo MJ, Lyons TW, Zhang C, Kelley C, Sassen R, Horita J, Cole DR (2004) Quantifying carbon sources in the formation of suthigenic carbonates at gas hydrate sites in the Gulf of Mexico. Chem Geo 205: 253-264.

Freytag JK, Girguis PR, Bergquist DC, Andras JP, Childress JJ, Fisher CR (2001) A paradox resolved: sulphide acquisition by roots of seep tubeworms sustains net chemoautotrophy. Proc Nat Acad Sci USA 98: 13408-13413.

Freytag JK (2003) Ecological physiology and biochemistry of sulfide acquisition by two hydrocarbon seep vestimentiferans, Lamellibrachia luymesi and Seepiophila jonesi. Ph.D. Thesis, Pennsylvania State University. 129 pp.

Gage JD, Tyler PA (1991) Deep-sea Biology: A Natural History of Organisms at the Deep-sea Floor. Cambridge University Press, Cambridge, UK. 504 pp.

176

Gaill F, Voss-Foucart MF, Shillito B, Goffinet G (1996) In situ dissolution of chitinous structures from hydrothermal vents. In: Advances in Chitin Sciences Vol. I., eds Domard A, Jeuniaux R, Muzarelli R, Roberts G, Andre publication, Lyon, France, pp. 143-148.

Gardiner SL, McMullin E, Fisher CR (2001) Seepiophila jonesi, a new genus and species of vestimentiferan tube worm (Annelida: Pogonophora) from hydrocarbon seep communities in the Gulf of Mexico. Proc Biol Soc Wash 114: 694-707.

Gardiner SL, Hourdez S (2003) On the occurrence of the vestimentiferan tube worm Lamellibrachia luymesi van der Land and Norrevang, 1975 (Annelida: Pogonophora) in hydrocarbon seep communities in the Gulf of Mexico. Proc Biol Soc Wash 116: 380-394.

George RY, Menzies RJ (1972) Deep-sea faunal zonation of benthos along Beaufort- Bermuda transect in the north-western Atlantic. Proc Roy Soc Edinb (B) 73: 183- 194.

Govenar BW, Bergquist DC, Urcuyo IA, Eckner JT, Fisher CR (2002) Three Ridgeia piscesae assemblages from a single Juan de Fuca Ridge sulphide edifice: structurally different and functionaly similar. Cah Biol Mar 43: 247-252.

Grieshaber MK, Volkel S (1998) Animal adaptations for tolerance and exploitation of poisonous sulfide. Ann Rev Physiol 60: 33-53.

Grimm V (1999) Ten years of individual-based modeling in ecology: what have we learned and what could we learn in the future? Ecol Model 115: 129-148.

Gros O, Duplessis MR, Felbeck H (1999) Embryonic development and endosymbiont transmission mode in the symbiotic clam Lucinoma aequizonata (Bivalvia: Lucinidae). Invertebr Reprod Dev 36: 93-103.

177

Girguis PR, Childress JJ, Freytag JK, Klose K, Stuber R (2002) Effects of metabolite uptake on proton-equivalent elimination by two species of deep-sea vestimentiferan tubeworm, Riftia pachyptila and Lamellibrachia cf luymesi: proton elimination is a necessary adaptation to sulfide-oxidizing chemoautotrophic symbionts. J Exp Biol 205: 3055-3066.

Gustafson RG, Turner RD, Lutz RA, Vrijenhoek RC (1998) A new genus and five new species of mussels (Bivalvia, Mytilidae) from deep-sea sulfide/hydrocarbon seeps in the Gulf of Mexico. Malacologia 40: 63-112.

Gutierrez JL, Jones CG, Strayer DL, Irbane OO (2003) Mollusks as ecosystem engineers: the role of shell production in aquatic habitats. Oikos 101: 79-90.

Hacker SD, Gaines SD (1997) Some implications of direct positive interactions for community species diversity. Ecology 78: 1990-2003.

Haedrich RL, Rowe GT (1977) Megafaunal biomass in the deep sea. Nature 269: 141- 142.

Harris LG, Ebeling AW, Laur DR, Rowley RJ (1984) Community recovery after storm damage: A case of facilitation in primary succession. Science 224: 1336-1338.

Head IM, Jones DM, Larter SR (2003) Biological activity in the deep subsurface and the origin of heavy oil. Nature 426: 344-352.

Henry P, Lallemant S, Nakamura K, Tsunogai U, Mazzotti S, Kobayashi K (2002) Surface expression of fluid venting at the toe of the Nankai wedge and implications for flow paths. Mar Geol 187: 119-143.

Hessler RR, Smithey WM, Boudrias MA, Keller CH, Lutz RA, Childress JJ (1988) Temporal change in megafauna at the Rose Garden hydrothermal vent (Galapagos Rift, eastern tropical Pacific). Deep-Sea Res 35: 1681-1709.

178

Hourdez S, Weber RE, Green BN, Kenney JM, Fisher CR (2002) Resipatory adaptations in a deep-sea orbiniid polychaete from Gulf of Mexico brine pool NR1: metabolic rates and hemoglobin structure/function relationships. J Exp Biol 205: 1669-1681.

Humphris CC (1979) Salt movement on continental slope, northern Gulf of Mexico. Am Assoc Pet Geol Bull 63: 782-798.

Hunter CM, Moller H, Fletcher D (2000) Parameter uncertainty and elasticity analyses of a population model: setting research priorities for shearwaters. Ecol Model 134: 299-323.

Iversen N, Jorgensen BB (1993) Diffusion coefficients of sulphate and methane in marine sediments: Influence of porosity. Geochim Cosmochim Acta 57: 571-587.

Johnson, KS, Childress, JJ, Hessler, RR, Sakamoto-Arnold, CM, Beehler, CL (1988) Chemical and biological interactions in the Rose Garden hydrothermal vent field, Galapagos Spreading Center. Deep-sea Res 35: 1723-1744.

Jones CG, Lawton JH, Shachak M (1994) Organisms as ecosystem engineers. Oikos 69: 373-386.

Jones ML (1985) On the vestimentifera, new phylum: six new species, and other taxa, from hydrothermal vents and elsewhere. Bull Biol Soc Wash 6: 117-158.

Joye SB, Boetius A, Orcutt BN, Montoya JP, Schulz HN, Erickson MJ, Lugo SK (2004) The anaerobic oxidation of methane and sulfate reduction in sediments from Gulf of Mexico cold seeps. Chem Geol 205: 219-238.

Julian D, Gaill F, Wood E, Arp AJ, Fisher CR (1999) Roots as a site of hydrogen sulphide uptake in the hydrocarbon seep vestimentiferan Lamellibrachia sp. J Exp Biol 202: 2245-2257.

179

Jumars PA, Gallagher ED (1982) Deep-sea community structure: Three plays on the benthic proscenium. In: Ernst WE, Morin JG (eds) The Environment of the Deep- sea. Rubey Colloquium, Los Angeles, CA. pp. 217-255.

Juniper SK, Sibuet M (1987) Cold seep benthic communities in Japan subduction zones: Spatial organization, trophic strategies and evidence for temporal evolution. Mar Ecol Prog Ser 40: 115-126.

Kennicutt MC, Brooks JM, Bidigare RR, Fay RR, Wade TL, McDonald TJ (1985) Vent- type taxa in a hydrocarbon seep region on the Louisiana slope. Nature 317: 351- 353.

Kennicutt MC, Brooks JM, Bidigare RR, Denoux GJ (1988) Gulf of Mexico hydrocarbon seep communities I. Regional distribution of hydrocarbon seepage and associated fauna. Deep-sea Res 35: 1639-1651.

Kennicutt MC, Brooks JM, Burke RA (1989) Hydrocarbon seepage, gas hydrates, and authigenic carbonate in the Northwestern Gulf of Mexico. In: Proceedings of the Offshore Technology Conference 5954, Houston, Texas. May 1-4, 1989. pp. 649- 654.

Kennicutt MC, McDonald TJ, Comet PA, Denoux GJ, Brooks JM (1992) The origins of petroleum in the northern Gulf of Mexico. Geochim Cosmochim Acta 56: 1259- 1280.

Kennicutt MC II, Schroeder WW, Brooks JM (1995) Temporal and spatial variations in sediment characteristics on the Mississippi-Alabama continental shelf. Cont Shelf Res 15: 1-19.

180

Kicklighter CE, Fisher CR, Hay ME (2004) Chemical defense of hydrothermal vent and hydrocarbon seep organisms: a preliminary assessment using shallow-water consumers. Mar Ecol Prog Ser 275: 11-19.

Lampitt RS, Billett DSM, Rice AL (1986) Biomass of the invertebrate megabenthos from 500 to 4100 m in the Northeast Atlantic Ocean. Mar Biol 93: 69-81.

Levin LA, Ziebis W, Mendoza GF, Growney VA, Tryon MD, et al. (2003) Spatial heterogeneity of macrofauna at northern California methane seeps: influence of sulfide concentration and fluid flow. Mar Ecol Prog Ser 256: 123-139.

Lill JT, Marquis RJ (2003) Ecosystem engineering by caterpillars increases insect herbivore diversity on white oak. Ecology 84: 682-690.

Lin S, Morse JW (1991) Sulfate reduction and iron sulfide mineral formation in Gulf of Mexico anoxic sediments. Am J Sci 291: 55-89.

Linström J, Kokko H (2002) Cohort effect and population dynamics. Ecol Let 5: 338- 344.

Liro LM (1992) Distribution of shallow salt structures, lower slope of the northern Gulf of Mexico, USA. Mar Pet Geol 9: 433-451.

Lowe VPW (1969) Population dynamics of the red deer (Cervus elaphus L.) on Rhum. J. Anim Ecol 38: 425-457.

Luther GW III, Rozan TF, Taillefert M, Nuzzio DB, Di Meo C, Shank TM, Lutz RA, Cary SC (2001) Chemical speciation drives hydrothermal vent ecology. Nature 410: 813-816.

Lutz RA, Kennish MJ (1993) Ecology of deep-sea hydrothermal vent communities: a review. Rev Geophys 31: 211-242.

181

MacArthur RH, Wilson EO (1963) An equilibrium theory of insular zoogeography. Evolution 17: 373-387.

MacAvoy SE, Carney RS, Fisher CR, Macko SA (2002) Use of chemosynthetic biomass by large, mobile, benthic predators in the Gulf of Mexico. Mar Ecol Prog Ser 225: 65-78.

MacAvoy SE , Macko SA, Carney RS (in press) Links between chemosynthetic production and mobile predators on the Louisiana continental slope: Stable carbon isotopes of specific fatty acids. Chem Geol 201: 229-239.

MacDonald IR, Boland GS, Baker JS, Brooks JM, Kennicutt MC, Bidigare RR (1989) Gulf of Mexico hydrocarbon seep communities. II. spatial distribution of seep organisms and hydrocarbons at Bush Hill. Mar Biol 101: 235-247.

MacDonald IR, Guinasso NL, Reilly JF, Brooks JM, Callender WR, Gabrielle SG (1990a) Gulf of Mexico seep communities: VI. Patterns in community structure and habitat. Geo-Mar Let 10: 244-252.

MacDonald IR, Reilly JF, Guinasso NL, Brooks JM, Carney RS, Bryant WA, Bright TJ (1990b) Chemosynthetic mussels at a brine-filled pockmark in the northern Gulf of Mexico. Science 248:1096-1099.

MacDonald IR, Guinasso NL, Ackleson SG, Amos JF, Duckworth R, et al. (1993) Natural oil slicks in the Gulf of Mexico are visible from space. J Geophys Res 98: 16351-16364.

MacDonald IR, Sager WW, Peccini MB (2003) Gas hydrate and chemosynthetic biota in mounded bathymetry at mid-slope hydrocarbon seeps: Northern Gulf of Mexico Mar Geol 198: 133-158.

182

MacDonald IR, Bohrmann G, Escobar E, Abegg F, Blanchon P, Blinova V, Brükmann W, Drews M, Eisenhauer A, Han X, Heeschen K, Meier F, Mortera C, Naehr T, Orcutt B, Bernard B, Brooks J, de Faragó (2004) Asphalt volcanism and chemosynthetic life in the Campeche Knolls, Gulf of Mexico. Science 304: 999- 1002.

Mana Capelli S, Busalmen JP, de Sanchez SR (2001) Hydrocarbon bioremediation of a mineral-base contaminated waste from crude oil extraction by indigenous bacteria. Int Biodet Biodeg 47: 233-238.

Maynard Smith J, Slatkin M (1972) The stability of predator-prey systems. Ecology 54: 384-391.

McGookey DP (1975) Gulf coast Cenozoic sediments and structure; An excellent example of extra-continental sedimentation. Trans Gulf Coast Assoc Geol Soc 25: 104-120.

McMullin ER (2003) Phylogeography of deep sea vestimentiferans and a population genetics study of two species Lamellibrachia luymesi and Seepiophila jonesi. Ph.D. Thesis, Pennsylvania State University. 122 pp.

McMullin ER, Hourdez S, Schaeffer SW, Fisher CR (2003) Phylogeny and biogeography of deep sea vestimentiferan tubeworms and their bacterial symbionts. Symbiosis 34: 1-41.

Milkov AV, Sassen R (2003) Preliminary assessment of resources and economic potential of individual gas hydrate accumulations in the Gulf of Mexico continental slope. Mar Pet Geo 20: 111-128.

Mills MA, Bonner JS, McDonald TJ, Page CA, Autenrieth RL (2003) Intrinsic bioremediation of a petroleum-impacted wetland. Mar Pollut Bull 46: 887-899.

183

Morse JW, Arvidson RS, Joye S (2002) Inorganic biogeochemistry of cold seep sediments In: MacDonald IR, editor. Stability and change in chemosynthetic communities, northern Gulf of Mexico. Report for U.S. Department of Interior, Minerals Management Service, Contract #1435-01-96-CT-30813, pp. 9-1 to 9-93.

Mortensen PB, Hovland M, Brattegard T, Farestveit R (1995) Deep water bioherms of the scleractinian coral Lophelia pertusa (L.) at 64 degrees N on the Norwegian shelf: structure and associated fauna. Sarsia 80: 145-158.

Mullineaux LS, Peterson CH, Micheli F, Mills SW (2003) Successional mechanism varies along a gradient in hydrothermal fluid flux at deep-sea vents. Ecol Monogr 73: 523-542.

Naiman RJ, Melillo JM, Hobbie JE (1986) Ecosystem alteration of boreal forest streams by beaver (Castor canadensis ). Ecology 67: 1254-1269.

Naiman RJ, Jonston CA, Kelley JC (1988) Alteration of North American streams by beaver. Bioscience 38: 753-761.

Nelson DC, Fisher CR (1995) Chemoautotrophic and methanotrophic endosymbiotic bacteria at deep-sea vents and seeps. In: Microbiology of Deep-Sea Hydrothermal Vents. ed Karl, D.M. CRC, Boca-Raton, Florida. pp. 125-167.

Nelson K, Fisher CR (2000) Absence of cospeciation in deep-sea vestimentiferan tube worms and their bacterial endosymbionts. Symbiosis 28: 1-15.

Newton PCD, Clark H, Edwards GR, Ross DJ (2001) Experimental confirmation of ecosystem model predictions comparing transient and equilibrium plant responses to elevated atmospheric CO2. Ecol Let 4: 344-347.

Odum EP (1969) The strategy of ecosystem development. Science 164: 262-270.

184

Olu K, Lance S, Sibuet M, Henry P, Fiala-Medioni A, Dinet A (1997) Cold seep communities as indicators of fluid expression patterns through mud volcanoes seaward of the Barbados accretionary prism. Deep-Sea Res 44: 811-841.

Page HM, Fisher CR, Childress JJ (1990) Role of filter-feeding in the nutritional biology of a deep-sea mussel with methanotrophic symbionts. Mar Biol 104: 251-257.

Pascual M, Roy M, Franc A (2002) Simple temporal models for ecological systems with complex spatial patterns. Ecol Let 5: 412-419.

Paull CK, Hecker B, Commeau R, Freeman-Lynde RP, Neumann C, Corso WP, Golubic S, Hook JE, Sikes E, Curray J (1984) Biological communities at the Florida Escarpment resemble hydrothermal vent taxa. Science 226: 964-967.

Peacock SM (1991) Numerical simulation of subduction zone pressure-temperature-time paths: Constraints on fluid production and arc magmatism. Phil Trans R Soc Lond Ser A 335: 341-353.

Pequegnat WE, Gallaway BJ, Pequegnat LH (1990) Aspects of the ecology of the deep- water fauna of the Gulf of Mexico. Amer Zool 30: 45-64.

Pilson MEQ (1998) An introduction to the chemistry of the sea. Upper Saddle River, NJ: Prentice Hall. 431 pp.

Pindell J (1985) Alleghenian reconstruction and subsequent evolution of the Gulf of Mexico, Bahamas, and proto-Caribbean Sea. Tectonics 4: 1-39.

Prior DB, Doyle EH, Kaluza MJ (1989) Evidence for sediment eruption on deep sea floor, Gulf of Mexico. Science 243: 517-519.

Redfield AC (1958) The biological control of chemical factors in the environment. Amer Sci 46: 205-221.

185

Ricciardi A, Bourget E (1998) Weight to weight conversion factors for marine benthic invertebrates. Mar Ecol Prog Ser 163: 245-251.

Roberts HH, Neurauter TW (1990) Direct observations of a large active mud vent on the Louisiana continental slope. Am Assoc Petrol Geol Bull 74: 1508

Roberts HH, Aharon P (1994) Hydrocarbon-derived carbonate buildups of the northern Gulf of Mexico continental slope: a review of submersible investigations. Geo- Mar Let 14: 135-148.

Röling WFM, Head IM, Larter SR (2003) The microbiology of hydrpcarbon degradation in subsurface petroleum reservoirs: perspectives and prospects. Res Microbiol 154: 321-328.

Rouse GW (2001) A cladistic analysis of Siboglinidae Caullery, 1914 (Polychaeta, Annelida): formerly the phyla Pogonophora and Vestimentifera. Zool J Lin Soc 132: 55-80.

Rosenfeld J, Pindell J (2003) Early Paleogene isolation of the Gulf of Mexico from the world’s oceans? Implications for hydrocarbon exploration and eustasy. Bull. Hous. Geo. Soc. 46: 15-17.

Rudinicki MD, Elderfield H, Mottl MJ (2001) Pore fluid advection and reaction in sediments of the eastern flank, Juan de Fuca Ridge, 48°N. Earth Planet Sci Lett 187: 173-189.

Sahling H, Rickert D, Lee RW, Linke P, Suess E (2002) Macrofaunal community structure and sulfide flux at gas hydrate deposits from the Cascadia convergent margin, NE Pacific. Mar Ecol Prog Ser 231: 121-138.

186

Sager WW, MacDonald IR, Hou R (2003) Geophysical signatures of mud mounds at hydrocarbon seeps on the Louisiana continental slope, northern Gulf of Mexico. Mar Geo 198: 97-132.

Sale PF (1977) Maintenance of high diversity in coral reef fish communities. Am Nat 111: 337-359.

Sanders HL (1968) Marine benthic diversity: a comparative study. Am Nat 102: 243-282.

Sarrazin J, Juniper SK (1999) Biological characteristics of a hydrothermal edifice mosaic community. Mar Ecol Prog Ser 185: 1-19.

Sassen R, Brooks JM, MacDonald IR, Kennicutt MC, Guinasso NL, Requejo AG (1993a) Association of oil seeps and chemosynthetic communities with oil discoveries, upper continental slope, Gulf of Mexico. Trans GCAGS 43: 349-355.

Sassen R, Roberts HH, Aharon P, Larkin J, Chinn EW, Carney R (1993b) Chemosynthetic bacterial mats at cold hydrocarbon seeps, Gulf of Mexico continental slope. Org Geochem 20: 77-89.

Sassen R, MacDonald IR, Requejo AG, Guinasso NL, Kennicutt MC, et al. (1994a) Organic geochemistry of sediments from chemosynthetic communities, Gulf of Mexico slope. Geo-Mar Lett 14:110-119.

Sassen R, Cole GA, Drozd R, Roberts HH (1994b) Oligocene to Holocene hydrocarbon migration and salt-dome carbonates, northern Gulf of Mexico. Mar Pet Geo 11: 55-65.

Sassen R, Joye S, Sweet ST, DeFritas DA, Milkov AV, MacDonald IR (1999) Thermogenic gas hydrates and hydrocarbon gases in complex chemosynthetic communities, Gulf of Mexico continental slope. Org Geochem 30: 485-497.

187

Sassen R, Losh SL, Cathles L, Roberts HH, Whelan JK, Milkov AV, Sweet ST, DeFreitas DA (2001) Massive vein-filling gas hydrate: relation to ongoing gas migration from the deep subsurface in the Gulf of Mexico. Mar Pet Geol 18: 551- 560.

Sassen R, Milkov AV, Ozgul E, Roberts HH, Hunt JL, Beeunas MA, Chanton JP, DeFeitas DA, Sweet ST (2003) Gas venting and subsurface charge in the Green Canyon area, Gulf of Mexico continental slope: evidence of a deep bacterial methane source? Org. Geochem. 34: 1455-1464.

Saunders JA, Thomas RC (1996) Origin of ‘exotic’ minerals in Mississippi salt dome cap rocks: results of reaction-path modeling. Appl Geochem 11: 667-676.

Schmidt-Nielsen K (1975) Animal physiology: adaptation and environment. Cambridge University Press, New York. 607 pp.

Schmitz OJ (2000) Combining field experiments and individual-based modeling to identify the dynamically relevant organizational scale in a field system. Oikos 89: 471-484.

Schroeder WW (2002) Observations of Lophelia pertusa and the surficial geology at a deep-water site in the northeastern Gulf of Mexico. Hydrobiologia 471: 29-33.

Scott KM, Fisher CR (1995) Physiological ecology of sulfide metabolism in hydrothermal vent and cold seep vesicomyid clams and vestimentiferan tube worms. Amer Zool 35:102-111.

Shea K, Wolf N, Mangel M. (in prep). Effects of density dependence on the detection of adult population trends from juvenile surveys in threatened coho salmon (Oncorhynchus kisutch).

188

Shepherd JG (1982) A versatile new stock-recruitment relationship for fisheries, and the construction of sustainable yield curves. J Const Int Explor Mer 40: 67-75.

Sibuet M, Olu K (1998) Biogeography, biodiversity and fluid dependence of deep-sea cold-seep communities at active and passive margins. Deep-Sea Res II 45: 517- 567.

Soto LA (1991) Faunal zonation of the deep-water brachyuran crabs in the Straits of Florida. Bull Mar Sci 49: 623-637.

Southward EC (1986) Gill symbionts in thyasirids and other bivalve molluscs. J Mar Biol Assoc UK 66: 889-914.

Stachowicz JJ (2001) The other side of the coin: positive effects of algal chemicals on animals. J Phycol 37: 46-56.

Stachowicz JJ, Hay ME (1999) Mutualism and coral persistence: The role of herbivore resistance to algal chemical defense Ecology 80: 2085-2101.

Streams ME, Fisher CR, Fiala-Medioni A (1997) Methanotrophic symbiont location and fate of carbon incorporated from methane in a hydrocarbon seep mussel. Mar Biol 129: 465-476.

Stumm W, Morgan JJ (2000) Aquatic Chemistry. New York: John Wiley & Sons, Inc. 1022 pp.

Sturges W, Evans JC (1983) On the variability of the loop current in the Gulf of Mexico. J Mar Res 41: 639-653.

Thiel V, Peckmann J, Richnow HH, Luth U, Reitner J, et al. (2001) Molecular signals for anaerobic methane oxidation in Black Sea seep carbonates and a microbial mat. Mar Chem 73: 97-112.

189

Treude T, Boetius A, Knittel K, Wallmann K, Jørgensen BB (2003) Anaerobic oxidation of methane above gas hydrates at hydrate ridge, NE Pacific Ocean. Mar Ecol Prog Ser 264: 1-14.

Tsurumi M, Tunnicliffe V (2001) Characteristics of a hydrothermal vent assemblage on a volcanically active segment of Juan de Fuca Ridge, northeast Pacific. Can J Fish Aquat Sci 58: 530-542.

Tunnicliffe V (1991) The biology of hydrothermal vents: ecology and evolution. Ann Rev Oceanog Mar Biol 29: 319-407.

Turnipseed M, Knick KE, Lipcius RN, Dreyer J, Van Dover CL (2003) Diversity in mussel beds at deep-sea hydrothermal vents and cold seeps. Ecol Let 6: 518-523.

Tyler PA, Young CM (1999) Reproduction and dispersal at vents and cold seeps. J Mar Biol Ass UK 79: 193-208.

Urcuyo IA, Massoth GJ, Julian D, Fisher CR (2003) Habitat, growth and physiological ecology of a basaltic community of Ridgeia piscesae from the Juan de Fuca Ridge. Deep-sea Res 50: 763-780. van der Heijden MGA, Klironomos JN, Ursic M, Moutoglis P, Streitwolf-Engel R, et al. (1998) Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396: 69-72.

Van Dover CL (2000) The ecology of deep-sea hydrothermal vents. Princeton University Press, Princeton, NJ. 424 pp.

Van Dover CL, Aharon P, Bernhard JM, Caylor E, Doerries M, Flickinger W, Gilhooly W, Goffredi SK, Knick KE, Macko SA, Rapoport S, Raulfs EC, Ruppel C, Salerno JL, Seitz RD, Sen Gupta BK, Shank T, Turnipseed M, Vrijenhoek R

190

(2003) Blake Ridge methane seeps: characterization of a soft-sediment, chemosynthetically based ecosystem. Deep-sea Res 50: 281-300.

Wallman K, Linke P, Suess E, Bohrmann G, Sahling H, et al. (1997) Quantifying fluid flow, solute mixing, and biogeochemical turnover at cold vents of the eastern Aleutian subduction zone. Geochim Cosmochim Acta 61: 5209-5219.

Welsh SE, Inoue M (2000) Loop current rings and the deep circulation in the Gulf of Mexico. J Geophys Res 105: 16951-16959.

Wicker AD, Detling JK (1988) Ecological consequences of prairie dog disturbance. BioSci 38: 778-785.

Wiedicke M, Sahling H, Delisle G, Faber E, Neben S, et al. (2002) Characteristics of an active vent in the fore-arc basin of the sundra Arc, Indonesia. Mar Geol 184: 121- 141.

Wishner KF, Ashjian CJ, Gelfman C, Gowing MM, Kann L, Levin LA, Mullineaux LS, Saltzman J (1995) Pelagic and benthic ecology of the lower interface of the Eastern Tropical Pacific oxygen minimum zone. Deep-sea Res 42: 93-115.

Won YJ, Hallam SJ, O’Mullan GD, Pan IL, Buck KR, Vrijenhoek RC (2003) Environmental acquisition of thiotrophic endosymbionts by deep-sea mussels of the genus Bathymodiolus. Appl Environ Microbiol 69: 6785-6792.

Wright JP, Jones CG, Flecker AS (2002) An ecosystem engineer, the beaver, increases species richness at the landscape scale. Oecologia 132: 96-101.

Young CM, Vázquez E, Metaxas A, Tyler PA (1996) Embryology of vestimentiferan tube worms from deep-sea methane/sulphide seeps. Nature 381: 514-516.

191

Zande JM (1999) An ascomycete commensal on the gills of Bathynerita naticoidea, the dominant gastropod at Gulf of Mexico hydrocarbon seeps. Invert Biol 118: 57-62.

Zar JH (1984) Biostatistical analysis. Prentice Hall, Englewood Cliffs, New Jersey.

Zhang CL, Li Y, Wall JD, Larsen L, Sassen R, et al. (2002) Lipid and carbon isotopic evidence of methane-oxidizing and sulphate-reducing bacteria in association with gas hydrates from the Gulf of Mexico. Geology 30: 239-242.

Zwolinski MD, Harris RF, Hickey WJ (2000) Microbial consortia involved in the anaerobic degradation of hydrocarbons. Biodegradation 11: 141-158.

Appendix: Abundance (#) or presence (P) of associated fauna. Chapter in which community data from each collection appears also noted. Central site collections arranged in order of increasing age.

chapter 6 4,5,6 4,5,6 5,6 5,6 4,5,6 4,5,6 6 4,6 5,6 4,6 4,5,6 4,5,6 4,6 5,6 6 6 4,5,6 4,5,6 4,5,6 4,5,6 6 Higher taxon Central Sites (Green Canyon) Species name bh2 1h 1g 4a 3b 1d 2a bh3 1b 3c 1c 2b 2c 1e 4b 3a gc3 1i 1a 1j 1f gc4 Porifera Ectomyxilla methanophila P P P P P P P P P P P Cnidaria actiniaria sp1 1 actiniaria sp2 actiniaria sp3 Callogorgia americanus P cerianthid 1 hydroid spp. P P P P P P P P P P P P P Phalangopora sp. P P P P P P P P P P P stolonifera spp. P P zoanthid P P P P P Platyhelminthes platyhelminthes spp. 1 1 Nemertea nemertean spp. 6 1 5 3 10 4 3 10 15 13 7 11 4 30 14 4 20 1 17 2 7 Sipuncula Phascolosoma turnerae 13 4 19 10 7 7 27 76 59 7 1 19 5 63 6 1 3 1 21 Annelida Polychaeta Ampharetidae Branchinotogluma sp. nov. 24 5 1 6 34 1 2 61 33 32 15 45 13 21 1 10 2 Branchinotogluma sp. nov.2 Branchipolynoe seepensis 42 Capitellidae sp.1 1 1 28 4 1 3 2 Capitellidae sp.2 Chaetopteridae

193

Cirratulidae Cirriformia sp. Ctenodrilidae 3 Dorivella sp. 1 Eteone sp. 12 1 2 6 21 6 1 67 1 3 5 18 1 Eulalia sp. 1 Eumida sunguinea 3 1 5 3 1 5 Eunicid 1 Eunice sp. nov. 1 14 3 6 1 5 6 18 7 2 2 Euratella sp. 13 1 Eurythoe sp. nov. 33 2 4 3 37 1 3 Euthalenessa sp. 1 Euthelepus sp. 1 2 1 Flabelligera sp. Genetyllis sp. 1 Glycera tesselata 2 2 1 2 8 3 3 2 4 5 6 3 13 1 1 Glycerid 1 Harmothoe sp. 4 17 2 2 6 18 2 28 10 27 18 21 10 11 19 10 55 11 27 2 1 1 Hesiocaeca methanicola 1 Lumbrineris tenuis 2 1 1 2 3 12 4 5 2 Lysidice ninetta 1 1 4 1 5 1 Maldanidae Mayella sp. 1 Methanoaricia dendrobranchiata 23 1 1 18 8 1 Nautilinelid Neoamphitrite sp. 5 5 3 1 2 2 Nephtys sp. 1 2 Nereis sp. nov 2 2 3 Nicomache sp. nov. 7 1 9 8 3 1 Nicomache sp.2 Orbiniid Paranaitis polynoides 1 1 3 5 Pholoe sp. 1 1 9 1

194

Phyllosyllis sp. 3 1 Pista sp. Pogonophoran Polynoidae Prionospio sp. Proscoloplos sp. 9 2 Protomystides sp. Pseudosyllides sp. 1 2 Questa sp. 1 Sabellastarte sp. 9 1 47 14 1 6 Sabellidae Scoloplos sp1 13 Scoloplos sp2 Spintheridae 2 Sthenelanella sp. 1 1 Synelmis sp. Terebellidae 1 1 2 3 1 Tharyx sp. 1 indet. Polychaete 6 1 1 Mollusca Aplacophora Aplacophora sp. 1 1 Aplacophora sp. 2 Polyplacophora Ischnochiton mexicanus 4 1 1 1 1 24 1 2 1 Lepidopleurida sp.1 1 3 3 15 7 4 16 3 Lepidopleurida sp.2 4 4 2 Gastropoda Bathynerita naticoidea 1536 121 164 1 207 1 6 38 62 2 2 Cancellaria rosewateri 6 2 10 13 1 3 2 1 Cantrainea macleani 25 2 5 2 2 Cataegis meroglypta 26 7 1 2 2 6 2 6 2 2 Cocculinidae

195

Diodora tanneri 1 2 2 2 1 3 2 2 5 7 Eosipho canetae 52 22 24 2 3 46 1 10 Fucaria sp. Gymnobella sp.nov. 3 1 Iothia sp.nov. 1 1 4 6 1 1 5 1 4 Lepetodrilus sp. Nudibranch Odostomia sp.nov. 2 15 1 Paraleptopsis sp. Phymorhyncus sp. Provanna sculpta 1074 336 79 23 61 1 330 6 528 7 4 8 5 1 2 1 Provannidae Turbonilla sp.nov. 2 4 3 Bivalvia Acesta bulli 1 1 8 Bathymodiolus brooksii Bathymodiolus childressi 172 39 18 10 20 4 1 17 1 4 2 4 1 Lucinoma sp. 2 6 1 2 5 2 2 1 Lucinidae Pecten sp. 1 Tamu fisheri 4 3 2 67 Thyasirid spp. 2 2 1 1 Vesicomyidae 2 5 1 1 5 1 3 1 3 1 indet. Bivalve 2 Arthropoda Crustacea Alvinocaris muricola Alvinocaris stactophila 279 30 23 1 306 27 259 377 42 44 9 52 65 6 3 28 47 4 21 2 2 Ampeliscid sp. 3 2 4 3 76 17 4 5 12 1 14 2 22 2 59 9 2 Amphipod sp.1 Amphipod sp.2 Amphipod sp.3 Atelecyclidae sp. 1 1 1 1 3 5 2 2 2 1 2 7 3 2 1

196

Bathynectes longispinus 1 Calaxius sp. 2 2 1 Costatoverruca floridana 5 3 1 1 1 1 6 2 3 2 Eusirid sp.1 1 2 Eusirid sp.2 1 1 Hyperiidae 1 Isopod sp.1 Munidopsis sp. nov. 1 8 18 16 12 87 23 81 96 3 30 31 35 60 169 45 33 60 32 13 75 9 Munidopsis sp. nov. 2 1 7 7 3 1 32 1 16 10 10 5 4 5 Munidopsis sp.3 Nibilia sp. 1 1 1 1 Periclimenes sp. 1 14 2 5 69 29 16 4 36 16 110 19 3 25 28 44 36 6 4 Phippsia sp.1 1 2 3 6 12 11 3 8 66 1 2 12 6 6 5 Phippsia sp.2 Phippsia sp. 3 Platyscelidae 1 Plesionika sp. 1 Rochinia crassa 1 1 Rochina tanneri 2 1 3 3 13 3 1 1 2 2 Stenopus sp. 1 Vibiliidae 1 Larval shrimp 9 2 Chelicerata Pycnogonida Anoplodactylus sp. 3 3 1 1 3 Collosendeis sp. Pentacolossendeis sp. 3 1 3 Entoprocta Bryozoan Sp. Echinodermata Holothuroidea Holothuroidea Sp. 1 Ophiuroidea

197

Amphioplus sp. Ophioneridae sp. nov 1 Ophiurocten spinilimbatum Asteroidea Sclerasterias tanneri 1 2 1 2 1 1 Chordata Agnatha Eptatretus springeri 2 Osteichthyes Bellottia sp. 1 1 2 2 1 6 5 3 2 Chauliodis sloani 2 Diaphus dummerli 1 Diplacanthapoma brachysoma 1 Dysommina rugosa 1 Ilyophis brunneus 3 2 1 1 1 Ophichthus cruentifer 1

198

chapter 6 5,6 6 6 5,6 5,6 5,6 6 6 6 6 6 6 6 6 6 6 6 Western Sites Eastern Sites Mussel sites Lower Slope Sites 7a 7d 7c 7b 5a 6a 6b 9ma 9mb 8ma ac1 ac2 acm1 acm2 atm1 fe1 fem1 fem2 Porifera Ectomyxilla methanophila P P Cnidaria actiniaria sp1 P actiniaria sp2 P P P actiniaria sp3 P Callogorgia americanus P cerianthid hydroid spp. P P P P Phalangopora sp. P stolonifera spp. P zoanthid P P P P P Platyhelminthes platyhelminthes spp. 2 Nemertea nemertean spp. 19 2 19 13 3 1 Sipuncula Phascolosoma turnerae 7 1 3 5 95 8 9 8 13 2 Annelida Polychaeta Ampharetidae 14 Branchinotogluma sp. nov. 2 10 3 26 4 10 Branchinotogluma sp. nov.2 2 4 15 Branchipolynoe seepensis 3 Capitellidae sp.1 1 6 3 Capitellidae sp.2 10 3 Chaetopteridae 3 Cirratulidae 3

199

Cirriformia sp. 1 Ctenodrilidae Dorivella sp. Eteone sp. 1 Eulalia sp. Eumida sunguinea 1 Eunicid Eunice sp. nov. 2 Euratella sp. Eurythoe sp. nov. 4 3 Euthalenessa sp. 2 5 Euthelepus sp. Flabelligera sp. 2 22 Genetyllis sp. Glycera tesselata 5 2 2 6 1 Glycerid Harmothoe sp. 14 2 1 6 26 1 1 6 5 Hesiocaeca methanicola 1 Lumbrineris tenuis 1 6 Lysidice ninetta 1 1 Maldanidae 3 1 Mayella sp. Methanoaricia dendrobranchiata Nautilinelid 5 Neoamphitrite sp. 45 21 Nephtys sp. 2 Nereis sp. nov 1 2 30 4 Nicomache sp. nov. Nicomache sp.2 10 Orbiniid 9 Paranaitis polynoides 1 Pholoe sp. 2 Phyllosyllis sp.

200

Pista sp. 3 Pogonophoran 7 7 Polynoidae 1 1 2 Prionospio sp. 22 Proscoloplos sp. Protomystides sp. 143 61 112 Pseudosyllides sp. Questa sp. Sabellastarte sp. 1 15 Sabellidae 1 Scoloplos sp1 Scoloplos sp2 3 Spintheridae 1 Sthenelanella sp. Synelmis sp. 2 Terebellidae 10 5 Tharyx sp. indet. Polychaete 1 Mollusca Aplacophora Aplacophora sp. 1 Aplacophora sp. 2 3 Polyplacophora Ischnochiton mexicanus 1 4 7 17 13 Lepidopleurida sp.1 1 1 4 6 1 Lepidopleurida sp.2 1 1 3 2 1 2 3 Gastropoda Bathynerita naticoidea 5 3 7 17 4 Cancellaria rosewateri 1 5 Cantrainea macleani 70 3 1 117 Cataegis meroglypta 1 4 71 66 Cocculinidae 4 Diodora tanneri 1

201

Eosipho canetae 2 3 4 1 Fucaria sp. 24 609 170 Gymnobella sp.nov. Iothia sp.nov. 2 3 Lepetodrilus sp. 2 1 Nudibranch 1 Odostomia sp.nov. 2 3 Paraleptopsis sp. 1 27 30 83 Phymorhyncus sp. 1 3 Provanna sculpta 17 14 44 1 Provannidae 1 Turbonilla sp.nov. 1 1 7 Bivalvia Acesta bulli 29 Bathymodiolus brooksii 1 Bathymodiolus childressi 9 3 2 1 5 Lucinoma sp. 1 1 Lucinidae 2 1 Pecten sp. Tamu fisheri 11 Thyasirid spp. 2 5 Vesicomyidae 3 indet. Bivalve Arthropoda Crustacea Alvinocaris muricola 5 23 35 85 2 4 3 1 Alvinocaris stactophila 11 18 33 17 26 1 Ampeliscid sp. 7 Amphipod sp.1 2 Amphipod sp.2 2 Amphipod sp.3 1 31 2 Atelecyclidae sp. 1 1 14 1 12 Bathynectes longispinus

202

Calaxius sp. Costatoverruca floridana 5 Eusirid sp.1 Eusirid sp.2 Hyperiidae Isopod sp.1 4 3 Munidopsis sp. nov. 1 2 12 3 15 60 21 30 Munidopsis sp. nov. 2 2 Munidopsis sp.3 1 3 Nibilia sp. Periclimenes sp. 5 2 1 3 3 1 3 Phippsia sp.1 1 3 1 Phippsia sp.2 3 Phippsia sp. 3 8 2 Platyscelidae Plesionika sp. Rochinia crassa 1 Rochina tanneri Stenopus sp. Vibiliidae Larval shrimp Chelicerata Pycnogonida Anoplodactylus sp. Collosendeis sp. 1 Pentacolossendeis sp. Entoprocta Bryozoan Sp. Echinodermata Holothuroidea Holothuroidea Sp. 1 4 1 Ophiuroidea Amphioplus sp. 1

203

Ophioneridae sp. nov Ophiurocten spinilimbatum 28 2 604 25 661 232 Asteroidea Sclerasterias tanneri Chordata Agnatha Eptatretus springeri Osteichthyes Bellottia sp. 1 Chauliodis sloani Diaphus dummerli Diplacanthapoma brachysoma Dysommina rugosa Ilyophis brunneus Ophichthus cruentifer

VITA

Erik E. Cordes

EDUCATION Ph.D. in Biology, December 2004. Penn State University M.S. in Marine Science, April 1999. Moss Landing Marine Labs B.S. in Marine Science, August 1993. Southampton College

PROFESSIONAL EXPERIENCE Research Associate, Continental Shelf Associates, Sept. 2003 - Sept. 2006 Research Associate, Moss Landing Marine Labs, Aug. 1999 - Aug. 2000 Biological Consultant, ABA Consulting, July 1999 - June 2000 Museum Curator, Moss Landing Marine Labs, Sept. 1998 - June 1999 Research Assistant, Moss Landing Marine Labs, Mar. - Dec. 1995

TEACHING EXPERIENCE Adjunct Faculty, Hartnell College, Salinas CA Marine Biology - Spring 1998-2000 General Biology - Summer 1998-2000, Fall 1999 General Biology (lab only) - Fall 1997-1998, Spring 1998-2000 Teaching Assistant, Penn State University Populations and Communities, Spring 2003 Invertebrate Zoology, Fall 2001 Teaching Assistant, Moss Landing Marine Labs Invertebrate Zoology II, Spring 1997 Marine Ecology, Fall 1996

GRANTS AND AWARDS Grant recipient (co-P.I.): Minerals Management Service. August 2003. Characterization of Northern Gulf of Mexico Deepwater Hard Bottom Communities with Emphasis on Lophelia coral (3 years) NOAA Dr. Nancy Foster Scholarship. July 2003. stipend plus tuition (up to 4 years) Award of Recognition for poster presentation at ALSO meeting, February 2003 PSU Center for Environmental Chemistry and Geochemistry Fellowship, summer 2002 Penn State University Graduate Fellowship, August 2000 (1 year) Braddock Scholarship, August 2000 (2 years), Pennsylvania State University Grant recipient (contributing): U. Alaska Fairbanks, North Pacific Research Initiative. August 1999. Age and growth of two deep-sea, habitat forming octocorallian corals of the Gulf of Alaska and Bering Sea, with radiometric age validation and climate correlation Grant recipient (P.I.): Dr. Earl and Ethyl Meyers Marine Biological and Oceanographic Trust April 1999. Laboratory Investigations of the Reproductive Biology and Growth of Anthomastus ritteri (Octocorallia: Alcyonacea) from Monterey Bay Member Beta Beta Beta Biological Honors Society. March 1992 University Scholarship. September 1989 (4 years), Southampton College, Long Island University

RESEARCH CRUISE EXPERIENCE I have participated in 9 research cruises over the past 10 years on board the R/V Atlantis, R/V Seward Johnson I, R/V Seward Johnson II and R/V Laney Chouest. I have made a total of 23 submersible dives in the DSV Alvin, DSV Johnson Sea-link I and DSV Johnson Sea-link II. I have also made numerous day trips on board the R/V Point Sur and R/V Point Lobos with the ROV Ventana during my work at Moss Landing Marine Labs.