the Atlantic Ocean by the St Georges Channel and to the north by the North ... largest offshore disposal ground in Europe (Beddington and Kinloch, 2005;.
The benthic ecology of Beaufort’s Dyke
Alexander David Callaway BSc, MRes
School of Environmental Sciences, Faculty of Life and Health Sciences, University of Ulster
Submitted for the degree of Doctor of Philosophy
CONTENTS TITLE PAGE
CHAPTER 1. INTRODUCTION
1.1 Benthic ecology
1.2 Habitat mapping
1.3 Benthic habitat mapping
1.4 Holistic approach
1.5 Study area
1.5.1 The Irish Sea
1.5.2 Beaufort’s Dyke
1.7 Thesis structure
CHAPTER 2. 21 THE FORMATION AND EVOLUTION OF AN ISOLATED SUBMARINE VALLEY IN THE NORTH CHANNEL, IRISH SEA: AN INVESTIGATION OF BEAUFORT’S DYKE.
2.1 Introduction 2.1.1 Study area 2.2 Methodology 2.2.1 Geophysical data processing 188.8.131.52 MBES data processing
21 25 29 29 29
184.108.40.206 Sparker data processing 2.2.2 Ground-truthing data processing
220.127.116.11 Particle size analysis (PSA)
18.104.22.168 Video data processing
2.3 Results and interpretation 2.3.1 Seabed data
22.214.171.124 Bathymetric data
126.96.36.199 Particle size analysis (PSA)
188.8.131.52 Video data
2.3.2 Sub-surface data 2.4 Discussion
2.4.1 Formation Processes
2.4.2 Effects of the BIIS and palaeotides on Beaufort’s Dyke
2.4.4 Recent geology of Beaufort’s Dyke
2.4.5 Further implications
CHAPTER 3. 62 THE HYDRODYNAMIC ENVIRONMENT OF A SUBMARINE TUNNEL VALLEY AND ITS EFFECT ON BENTHIC EPIFAUNA: BEAUFORT’S DYKE, IRISH SEA. 3.1 Introduction
3.2.1 Data acquisition
184.108.40.206 Acoustic survey
220.127.116.11 Atmospheric and oceanographic data
18.104.22.168 Hydrodynamic model
22.214.171.124 2D model
126.96.36.199 3D model
188.8.131.52 Video data acquisition
184.108.40.206 Grab sampling
220.127.116.11 Particle size analysis (PSA)
3.2.2 Relationship between model-derived variables and sediment properties 3.3 Results
3.3.1 Acoustic Survey
3.3.2 Hydrodynamic model
18.104.22.168 Surface currents
22.214.171.124 Internal current speed
126.96.36.199 Horizontal current direction
188.8.131.52 Turbulent kinetic energy
3.3.3 Particle size analysis (PSA)
3.3.4 Video data analysis
3.3.5 Model-sediment relationships
3.4.1 Hydrodynamic model
3.4.2 Potential sediment dynamics
3.4.3 Hydrodynamic effects on epifauna
CHAPTER 4. HEAVY METAL CONTAMINATION OF BEAUFORT’S DYKE, NORTH CHANNEL, IRISH SEA: A LEGACY OF ORDNANCE DISPOSAL.
4.2.1 Data acquisition
184.108.40.206 Acoustic survey
220.127.116.11 Grab sampling
4.2.2 Hydrodynamic and particle transport model
18.104.22.168 3D model
22.214.171.124 Particle transport
4.2.3 Particle size analysis (PSA)
4.2.4 Heavy metal analysis
4.3.1 Heavy metal analysis
4.3.2 Comparison with BGS and SOAFED samples
4.3.3 Particle transport simulation
CHAPTER 5. BENTHIC HABITAT MAPPING FROM A SPECIES’ ECOLOGY PERSPECTIVE: A CASE STUDY OF BEAUFORT’S DYKE, NORTH CHANNEL, IRISH SEA.
5.2.1 Data Acquisition
5.2.2 MBES data processing
5.2.3 Ground-truthing data processing
126.96.36.199 Grab sample processing
188.8.131.52 Particle size analysis (PSA)
184.108.40.206 Heavy metal analysis
220.127.116.11 Statistical Analyses
5.2.4 Atmospheric and oceanographic data
5.2.5 Hydrodynamic model
5.2.6 Predictive species distribution modelling
5.3.1 Comparison with NMMP and North Channel Peaks sites
5.3.2 Beaufort’s Dyke results
5.3.3 Predicted species distributions
5.3.4 Probable assemblage distributions and habitat map
5.4.1 Comparison with neighbouring regions
5.4.2 Ecology of Beaufort’s Dyke
5.4.3 Predicted species distribution and habitat map
5.4.4 Further implications
CHAPTER 6. DISCUSSION
6.1.1 Age and evolution of Beaufort’s Dyke
6.1.2 The hydrodynamic model
6.1.3 Contamination from ordnance
6.1.4 Species distribution in the Dyke
6.2 Limitations of the study
6.3 Suggestions for further research
ACKNOWLEDGEMENTS I would like to thank my parents, Christina and Paul Callaway, for their unwavering support and encouragement.
I express my gratitude to my supervisors Craig Brown, Rory Quinn, Matthew Service and Sara Benetti for their academic guidance, editorial prowess and patience throughout this project.
I would like to thank the Environmental Sciences Research Institute (University of Ulster) and Department for Employment and Learning for financial support; the Agri-Food and Biosciences Institute for project support and heavy metal analysis; the Captain and crew of RV Corystes for their assistance and patience in data collection. Brian Stewart and his team for data collection and sea going experience. I also thank Dave Long (British Geological Survey) for seismic data acquisition and geological expertise and David Wallis (BGS) for seismic data acquisition. Seismic Micro Technology for providing Kingdom Suite to the School of Environmental Sciences at the University of Ulster and DHI for providing MIKE software. I also thank Richard Hartley (University of Plymouth) for carrying out particle size analysis and Irene Delagado-Fernandez for informative discussions on sediment transport plus Proudman Oceanographic Laboratory for providing Meteorological data. To those in the benthic ecology/geophysics basement, namely, Chris McGonigle, Henk Van Rein, Ruth Pletz and Annika Clements, I thank you for GIS, methodological and statistical advice, numerous beverages, discussions, digressions and experiences shared over the last three years. And Laura, thank you for counting hundreds of worms and generally putting up with me.
SUMMARY This work represents an holistic study of the benthic environment. This research comprises Quaternary, hydrodynamic modelling, contaminant particle transport simulation and species-centric predictive habitat mapping. Understanding the formation processes of a Quaternary feature and the potential for modification of the seabed by prevailing currents creates a basis for comprehensive ecological studies of the benthos. Subsequently, the method described in this study offers a new strategy for marine habitat mapping. The formation mechanism of Beaufort’s Dyke explains the geomorphology and substrata type of the area which are important variables for faunal assemblages and detailed hydrodynamic information from simulations increases the strength of subsequent ecological investigation. Using simulation results, differences in epifaunal community structure and abundance between sample regions can be attributed to enhancement of colonisation by near-bed currents and life history response to hydrodynamics rather than substratum composition alone. Also, the potential for modification of bedforms by prevailing currents can be tested by comparing hydrodynamic and sedimentological data. An observed increase in the concentration of heavy metals within Beaufort’s Dyke sediments may be the first evidence of the legacy of ordnance disposal around the Dyke. Further increase in contaminant levels within Beaufort’s Dyke sediments will negatively impact resident fauna and the ecology of the region. Particle transport simulations demonstrate that dispersal of heavy metals from Beaufort’s Dyke is possible and that disposed ordnance may also contribute to pollution of surrounding areas. Habitat conservation is often initiated because of resident fauna that are deemed vulnerable but these fauna are subsequently omitted in
the map creation process. By utilising hydrodynamic, acoustic and faunal data to objectively determine assemblage relationships and combining faunal distribution data to produce species-centric habitat maps, management of vulnerable species can be based upon the target species, its associated community and habitat rather than abiotic surrogates.
“I hereby declare that with effect from the date on which the thesis is deposited in the Library of the University of Ulster, I permit the Librarian of the University to allow the thesis to be copied in whole or in part without reference to me on the understanding that such authority applies to the provision of single copies made for study purposes or for inclusion within the stock of another library. This restriction does not apply to the British Library Thesis Service (which is permitted to copy the thesis on demand for loan or sale under the terms of a separate agreement) nor to the copying or publication of the title and abstract of the thesis. IT IS A CONDITION OF USE OF THIS THESIS THAT ANYONE WHO CONSULTS IT MUST RECOGNISE THAT THE COPYRIGHT RESTS WITH THE AUTHOR AND THAT NO QUOTATION FROM THE THESIS AND NO INFORMATION DERIVED FROM IT MAY BE PUBLISHED UNLESS THE SOURCE IS PROPERLY ACKNOWLEDGED”.
CHAPTER 1: INTRODUCTION
1.1 Benthic ecology
Benthic ecology was introduced as a discipline in the early 20th Century when Petersen (1914) described assemblages of benthic species’ that occurred in areas of varying physical conditions. This approach was based on describing sedimentological qualities and the associated fauna found in samples along transects (Gray and Elliott, 2009). However, ecology is defined as the study of interactions between organisms and their environment which demands more information than sediment qualities and resident fauna. In benthic ecosystems the physical environment is structured by available substrata which are sorted by hydrodynamic and geological processes. For example, sand waves can only occur in areas proximal to where sand has been made available through erosive processes and subsequently transported. Sand waves are then structured by tide and wave oscillations although the influence of these forces is reduced as water depth increases (Uehara et al., 2006; Van Landeghem et al., 2009). The hydrodynamic environment is dependent on the lunar cycle and prevailing winds which influence the water depth, dominant flow direction and velocity and the geological environment which constrains aquatic basins. The chemical environment is determined by anthropogenic inputs, temperature, depth, erosion and rain fall which can all influence the acidity, salinity and nutrient availability in the water column and sediments. Finally, the biological environment is dependent on all physical aspects of the environment but also on various biological processes. Without water, substrata and nutrients, marine
organisms cannot exist. The acidity, salinity, temperature, and current regime determine the surrounding conditions for all marine organisms whilst substrata directly influence the benthos. Marine organisms also influence the distribution of each other through mechanisms such as symbiosis (corals), competition for resources (space, food, sexual), resource provision (commensalism) and predation (Nybakken, 1996). It is the lack of holistic information on these interactions and the potential impact from anthropogenic activities (aggregate removal, fishing, pollution, climate change) that formulates the basis for modern investigations of marine ecology. One of the most rapidly developing disciplines in marine ecological investigation is benthic habitat mapping.
Figure 1.1. Interrelationship of abiotic and biotic environments. Arrows indicate direction of controlling influence.
1.2 Habitat mapping
What is a habitat? There are multiple definitions of habitat offered and adopted (Hall et al., 1997). For example, Speight and Henderson (2010) state that aquatic habitats are classified by depth and locality within the water column (i.e. neritic, benthic, etc) whereas Begon et al. (1996) define habitat as the place where an organism lives. These definitions are confusing. One states that habitat is a physical entity, the space on the Earth where biological organisms can exist whilst the other infers that a habitat is defined by the presence of biota. It is the latter definition that is in agreement with examples presented in this thesis. The reason for this is because, by definition, without biota a habitat is not a place where an organism lives, it is a place where an organism might live provided the conditions are suitable. However, this suitability is not evident until biota are observed in an area. For example, Mytilus edulis is a marine bivalve that can exist in both inter- and sub-tidal environments and take advantage of almost the entire continuum of exposures, salinities and substrata, fine mud being the exception (Tyler-Walters, 2008). However, M. edulis is not found on all available substrata indicating a limiting biological factor. This could be food supply or inter-specific competition. For example, on rocky shores, if the exposure time is too long then species with higher desiccation resistance, such as Cthalamus montagui, prosper where M. edulis cannot. Subtidally, predation from Nucella lapillus or Carcinus maenas may limit the population of M. edulis. Thus, the very presence of biota changes the area from a suite of physical conditions (e.g. rock, hydrothermal vent) to a habitat. It can be argued that only physical information is required to describe habitats. From substratum alone
rocky habitat, sand habitat and mud habitat are often described. These three broad categories support a range of species’ some of which are epilithic and will be expected on rock and others that are psammophilic and will be expected to be found in sand habitats. However, the true habitat of the organism is not simply rock or sand. It is the percentage contribution of various substrata, roughness characteristics, chemical conditions, temperature, hydrophysical regime and biological interaction that define species presence. A more fundamental problem with the non-biological physical entity view of habitat is that some organisms live within or upon other organisms (e.g. Helcion pellucidum on Laminaria digitata). This creates a paradox for the physical view of habitat where the habitat for the epibiota or parasite is actually a biological organism. To make the term more suitable for habitat mapping techniques, Kostylev et al. (2001) expand on the definition of habitat describing habitat as a spatially defined area where the physical, chemical and biological environment is distinctly different from the surrounding environment. This is the definition of habitat that is used from this point forward.
1.3 Benthic habitat mapping
Benthic habitat mapping is the process of combining physical data with biological data to produce a map of biotic distributions within a sample region (McCrea et al, 1999; Kostylev et al., 2003; Galparsoro et al., 2009; Brown et al., 2011a). In its most basic form habitat mapping involves correlating biotic variables such as community structure with environmental variables including
Figure 1.2. a) SBES mapping results (Freitas et al., 2003) and b) MBES mapping results (Diesing et al., 2009). The illustrations demonstrate the disparity in resolution and potential size of study area afforded by the two different acoustic systems. i.e. MBES more efficient in collecting large, continuous data sets than SBES.
depth, grain size, salinity and temperature from point samples (Cartes et al., 2007). This enables a description of the environment in which sampled communities exist although the lack of continuous physical data restricts the extrapolation of regional trends from observations. To overcome this problem the discipline has taken advantage of geoacoustic methods.
The use of single-beam echo-sounders (SBES), sidescan sonar (SSS) and acoustic ground-discrimination systems (AGDS) for collection of acoustic data enabled a greater density of seabed data to be collected than grab sample transects informed by charts measured with lead lines (Gray and Elliott, 2009). These data can be segmented either visually or using computer programmes such as QTC View (Frietas et al., 2003), which segments data based upon principal components analysis of the acoustic returns, and compared against ground-truth data to provide a more complete habitat description. However, the disparity between acoustic and ground truth data resolution coupled with the level of interpolation often implemented again reduces the confidence in describing regional trends. The advent and increased availability of highresolution multi-beam echo-sounders (MBES) greatly reduced the problem of acoustic data paucity (Figure 1.2). A recent review by Brown et al. (2011a) illustrates the emergence of MBES as the method of choice for acoustic remote sensing of the seabed. This is because of the ability of MBES systems to simultaneously collect coincident, high-resolution bathymetry and backscatter data. The increased availability of such comprehensive data sets has led to numerous methods of data interpretation although these methods can be categorised into three strategies:
(1) unsupervised classification of abiotic surrogates, e.g. Using programmes such as QTC Multiview to segment acoustic data by decibel ranges and using segmented maps a proxy for biological distributions (2) assemble first, predict later (unsupervised classification), e.g. as strategy 1but correlating in situ faunal samples with abiotic data to enable prediction of species occurrence and (3) predict first, assemble later (supervised classification), e.g. faunal distributions are predicted by combining in situ faunal samples with environmental data and correlating species presence records with environmental variables. Maps are then segmented based upon the strongest biotic/abiotic correlations (Brown et al., 2011a). Strategy 1 is widely used in geological studies (Brown et al., 2011b) and strategy 2 is the most popular in benthic habitat mapping (Brown et al., 2011a). More recently, marine investigations have been utilising comprehensive environmental data sets to inform species distribution models which provide more ecologically relevant information for use in habitat maps than previous investigations (Brown et al., 2011a). These methods are established in terrestrial ecology but considered emerging techniques in marine studies because of the difficulty in attaining total data coverage in the sublittoral (Brown et al., 2011a). Results of species distribution models will enable more habitat mapping investigations to take advantage of strategy 3 thus providing greater information for management of marine resources. It can be argued that advances in the technology and ability to discriminate differences in acoustic data for seabed classification resulted in habitat mapping becoming too reliant on the acoustic-sediment relationship. Several studies report significant correlations of backscatter with grain size (e.g. Goff et al., 2000; Lathrop et al., 2006; Collier and Brown, 2008) and sediment-biota
relationships have also been described (Hovland et al., 2002; Freitas et al., 2006; Cartes et al., 2007). These relationships and the availability of software to classify large datasets enabled broad-scale habitat maps to be described efficiently with minimal ground-truthing. This method works particularly well when describing the distribution of what are essentially monoculture assemblages. However, environmental variables extend beyond sedimentology and other controlling factors influence biota (Figure 1.1). Therefore, in more complex systems a more holistic approach to benthic habitat mapping is required.
1.4 Holistic approach
Using the definition of habitat described by Kostylev et al. (2001) it is necessary to implement multidisciplinary studies of the marine environment to understand the processes responsible for patterns observed in benthic habitats. Acoustic data can provide comprehensive coverage of the seabed from which variables such as depth, slope angle, curvature and roughness can be derived. These variables describe the structural environment of benthic habitats. Salinity influences biota and is obtained by direct sampling with a CTD cast (Kaiser et al., 2005). Temperature data are often available and collected from direct sampling techniques (CTD cast) or from remote sensing data (satellite imagery) (Connell and Gillanders, 2007). Hydrodynamic properties, such as current speed and turbulent kinetic energy (TKE), also influence the colonisation of benthic fauna and affect the supply of nutrients to an area as well as sorting the sediment upon which fauna reside. Pollution can also determine the distribution
Figure 1.3. Interaction between abiotic and biotic elements required for holistic ecological analysis. Coloured lines indicate interactions at different levels within a community. Black, blue and green lines indicate abiotic and biotic interactions, the red line indicates that all interactions are required for ecological analysis of a community.
of biota by acting as a disturbance and removing species from an area by causing premature mortality (Long et al., 1995). By integrating all of these environmental variables a more detailed picture of the factors that determine species distribution and thus habitat extents can be achieved (Figure 1.3). However, the availability of large geospatial data sets of environmental variables for benthic environment remains poor. Other sampling systems have not experienced the equivalent technological advances that acoustic systems
have or are not suited to broad scale remote sensing operations. Current meters and acoustic doppler current profilers are expensive to deploy and provide either point or line data, thus the level of interpolation required for complete coverage diminishes the accuracy of the data. The problem of resolution can be overcome with the use of numerical modelling techniques. Hydrodynamic simulation software can accurately recreate oceanographic conditions throughout the water column at a higher resolution than remote sensing techniques. This work demonstrates that these techniques provide an opportunity to reconcile oceanographic and acoustic data resolution to enable holistic ecological investigations. Recent benthic habitat investigations have taken advantage of habitat suitability models commonly used in terrestrial ecology (Galparsoro et al., 2009; Merckx et al., 2010; Huang et al., 2011). These methods allow users to test the contribution of multiple environmental variables against the occurrence of species and determine which factors contribute most to their distribution (Phillips et al., 2006). This is only possible because of the improved quality and coverage of environmental variables available for marine systems (Gogina and Zettler, 2010). Huang et al. (2011) tested the efficacy of 15 species distribution models at predicting sponge habitats when provided with 26 environmental variables and 1566 presence records. They concluded that machine learning models such as random forest decision tree, probabilistic neural network and MaxEnt performed best (Huang et al., 2011). This presents a step change in benthic habitat mapping. The utilisation of multiple environmental layers to predict species distributions offers an opportunity to map habitat extents based on the probabilistic distribution of
species contributing to benthic assemblages. There is also the potential to create habitat maps that not only illustrate the distribution of benthic species but also infer the ecological relationships within benthic habitats by combining probabilistic distributions with traditional community analysis techniques such as those offered within the PRIMER statistics package (Clarke and Warwick, 2001) (Figure 1.3). This in turn will provide better information for marine management agencies enabling more effective management and conservation of marine ecosystems.
1.5 Study Area
1.5.1 The Irish Sea
The Irish Sea is a semi-enclosed sea that is bounded by the Republic of Ireland, and the United Kingdom (Figure 1.4). To the south the Irish Sea is connected to the Atlantic Ocean by the St Georges Channel and to the north by the North Channel (Figure 1.4). The Irish Sea was a main conduit for the British and Irish Ice Sheet (BIIS) during the last glacial maximum and was deepened by the Irish Sea Ice Stream (Evans and Ó Cofaigh, 2003). This past glacial environment produced a large volume of glacigenic sands which now dominate the seabed substrata although significant post-glacial mud deposits also exist in some regions (Jackson et al., 1995). The Irish Sea has shallow basins in the east with depths