A seascape perspective for managing deep-sea habitats

A. Williams, R. Kloser, B. Barker, N. Bax and A. Butler
CSIRO Marine Research
PO Box 1538, Hobart, Tasmania, Australia, 7001
<[email protected]>

1. INTRODUCTION

Sustainable use of the deep seabed off southeastern Australia is presently a focus for marine planning agencies, conservation groups, fishery managers and user groups including the offshore fishing industry. An important stimulus for this focus is Australia’s Oceans Policy (Commonwealth of Australia 1998) which is being implemented through Regional Marine Plans (RMPs) (NOO 2003) that include a National Representative System of marine protected areas (ANZECC 1999) so as to encompass many areas of continental shelf and slope seabed. In addition, and largely independently, an expanded control of the offshore fisheries of the region’s South East Fishery (SEF) through spatial management is signalled by a range of fishery-specific management planning by the Australian Fisheries Management Authority through bycatch action plans, strategic fishery assessments and ecological risk assessments. Ecologically sustainable development in the SEF, including spatially based management concepts, is also the focus of conservation NGOs (e.g. Ward and Hegerl 2002), and ecologically-sound fishing practices are supported by the strategic plans of peak industry associations, including the trawl sector.

At present, however, the information needs of this spatial policy focus considerably exceed our existing knowledge of the large offshore seabed areas. This is particularly the case for the outer continental shelf and continental slope that support a considerable and expanding fishing effort, but which are largely unseen and their ecosystems poorly understood. In the vast area to be managed by the first RMP (> 2 000 000 km2), the two exceptions are a group of cinder cones (now the Tasmanian Seamounts Reserve) and an area of continental shelf (the Twofold Shelf bioregion) studied by Koslow et al. (2001) and Bax and Williams (2001) respectively. Nonetheless, those studies, in common with those undertaken elsewhere (e.g. Langton, Auster and Scheider 1995), demonstrated the multiple spatial scales at which seabed habitats and biodiversity exist, and therefore the multiple scales at which structures and functions of marine benthic ecosystems are organized. Understanding these patterns at the landscape scale is now recognized as essential to successfully managing natural resources for objectives such as biodiversity conservation and ecologically sustainable development (Simberloff 1998; Roff and Taylor 2000, ANZECC and BDAC 2001). Therefore, one of the needs for effective and integrated spatial planning on the deep seascape off SE Australia is that of identifying the spatial scales at which information is required.

2. METHODS AND DATA SOURCES FOR HABITAT CLASSIFICATION

Here we provide a multi-spatial scale perspective for habitat distribution on the upper continental slope - the seabed region bounded approximately by the 200 and 700 m isobaths - and review briefly the relevance of each scale to scientific survey (especially mapping), habitat use, and habitat management. We do this by describing a variety of seabed habitats that make up 150 km2 of a large terrace at the SE margin of the Big Horseshoe Canyon (the Big Horseshoe Canyon SE terrace) collected as part of a larger habitat mapping survey (Kloser, Williams and Butler 2001a, b). Summary details of habitats come from Williams et al. (2004). A framework for our multi-spatial scale classification of habitats is provided by a hierarchical scheme being adopted for spatial planning by the Regional Marine Plans (NOO 2003; Williams and Bax 2003a). A hierarchal classification of "habitats" is effectively used as a surrogate for the hierarchy of ecological units and processes. The scheme applied to the SER recognizes a series of nested, pseudo-spatial ‘Levels’ for the structure of habitats, each reflecting the influence of characteristics and processes acting at different scales (Table 1). It is mainly under development by V. Lyne and P. Last of CSIRO Marine Research, Hobart. In addition, explicit spatial scales for habitats are defined according to the scheme of Greene et al. (1999).

3. HABITAT SCALES AND LINKS TO SURVEY, USE AND MANAGEMENT (TABLE 1)

3.1 Provincial scales

Provinces are the first level in the classification scheme, and they divide Australia’s SE Region of more than 2 000 000 km² into large areas based on regional patterns in fauna (CSIRO Marine Research 2001) and physiography. Our study area falls within the easternmost Province 3 off the SE Australian continental margin. It is a Level 1 habitat of some 500 000 km2. Within the province, Biomes defined by major community types and physiography separate the continental slope from the adjacent continental shelf and continental rise at Level 2 in the classification scheme (Table 1). Off SE Australia, the upper continental slope is a 3 000 km long sinuous ribbon of seabed that averages only 7.2 km in width as it winds around the continental margin immediately seaward of the shelf break between depths of about 200 and 700 m.

Depth is the strongest environmental correlate of fish community structure in the deep temperate Australian marine environment (see references in Williams and Bax 2001b), and the southeastern upper slope is defined biologically as a Sub-biome at Level 2b (Table 1). It has a distinct demersal fish community that differs markedly to those at the adjacent shelf-break and the mid-slope (CSIRO Marine Research 2001, Last et al. 2005).

At the largest habitat scales, biogeographic provinces, biomes and sub-biomes provide the context to view the habitats of the Big Horseshoe Canyon SE terrace. Their attributes are the large scale environmental variables of latitude, depth and hydrology (at several scales) that correlate with the distributions of marine communities (biodiversity) and fishery resources (Bax and Williams 2001 and references therein). The Big Horseshoe Canyon SE terrace can therefore be visualized as making up part of a habitat restricted to the approximately 300-600 m depth zone on the upper slope in the eastern province of the SE region. Its communities include a suite of large benthic and benthopelagic fishes, including the commercially-exploited pink ling where it occurs at its peak population abundance, and is targeted by the offshore fishing fleet made up by trawlers and ‘non-trawl’ boats fishing with hook and line, gillnets and traps. As a result of the narrowness of this depth zone it has a relatively small area overall (11 250 km2), and a correspondingly small fraction of the South East Fishery (SEF) region, i.e. 5 percent of the 227 340 km2 of the area used for fishing outside coastal waters defined as from 3 nm from shore to 1 300 m depth.

3.2 Megahabitat scales (km to 10s of km and larger)

Geomorphic features at large megahabitat spatial scales and the biological communities they support are represented at Level 3, the next level in the habitat classification scheme as Major Biogeomorphological Units. The study area represents one of these units: the terrace being the western extent of sediment plain that extends eastwards. Additional Level 3 habitat units are represented by the other regions that bound the study area: the main arm of the Big Horseshoe Canyon that extends rapidly to 850 m depth to the west, the shelf break escarpment characterized by a series of slumps, scarps and steep slopes in the approximately 200-300 m depth range to the north, and the mid-slope (> 700 m depth) to the south. While submarine canyons are prominent features of the continental slope seabed off SE Australia (NOO 2003, p. 54), with over 100 primary or tributary canyons estimated to intersect the 300-600 m depth zone in the South East Fishery region, Big Horseshoe Canyon is distinctive in forming one major arm of Bass Canyon, the region’s largest canyon. Within the restricted upper slope habitat, the Big Horseshoe SE terrace is therefore an example of a habitat of 150 km2 existing in a location that is unique with respect to its topography and hydrodynamic climate.

TABLE 1
Benthic habitats from the Big Horseshoe southeast terrace classified in the hierarchical habitat classification scheme proposed for Australia’s marine environment

This scale represents the largest units of the continental shelf and slope that can be mapped cost-effectively by swath acoustics and ‘ground-truth’ sampling (i.e. over a period of days) (Kloser et al. 2002). Swath acoustics provide complete mapping coverage of the seabed (Exon and Hill 1999) and enables visualization at scales of 10 km as well as production of maps based on detailed bathymetry and seabed textures. This allows scientific sampling to be targeted at particular seabed features or textures (Kloser et al. 2002). Habitats at this scale may correspond to locally distinct ecosystems such as canyons that are defined by topography and, or, locally defined circulation, and may support enhanced productivity and biological aggregations e.g. Big Horseshoe Canyon (Bax and Williams 2001). As such, habitats at this scale are correlated with the general distribution of fishing grounds and the study area is an example of a large multi-sector fishing ground. Vulnerability may be assessable at this level based on knowledge of geological properties of habitats and impact studies made at finer scales. Collectively, these factors identify major geomorphological units at large megahabitat scales as the largest operational scale for managing anthopogenic habitat use.

'Acoustic facies’ (Kloser et al. 2002) form mosaics at smaller megahabitat scales (1 km-10s of km), and are the equivalent of Primary Biotopes, or Level 4 units, in the proposed Australian scheme (Table 1). Three types of acoustic facies form the Big Horseshoe Canyon SE terrace: (a) large areas of homogeneous flat seabed characterized by low multibeam reflectivity which make up the majority of the area (approximately 1012 of the 150 km2) and are interpreted a priori as ‘Soft’ substratum - sediments; (b) smaller interspersed heterogeneous areas characterized by relatively high reflectivity (six patches making up approximately 432 of 150 km2): interpreted a priori as ‘Hard’ substratum - i.e. consolidated material; and (c) a patch found on the western margin of the terrace of high acoustic reflectivity that occurs on a steep (to 150²) slope (~6 km2 of the 150 km2): interpreted a priori as ‘Rough’ substratum - consolidated material exposed on steeply sloping seabed.

During sampling with a video camera to observe these acoustic facies (see below), a total of 85 individuals of adult pink ling were observed; they were strongly associated with structured microhabitats provided by the rough habitat (microhabitats detailed below) and had approximately 30 times higher density of individuals on this primary biotope than the other two.

Primary biotopes, existing at megahabitat scales, make up the major geomorphological units and are the appropriate scale at which to understand habitat values, the interaction of users with the seascape, e.g. fishing effort and catch, and for scientists to direct scientific sampling of habitats (Bax and Williams 2001). Of particular importance is that photography and physical sampling confirmed that habitats at this level were successfully differentiated by multi-beam acoustics as their general distributions corresponded well to the a priori designation of ‘Soft’, ‘Hard’ and ‘Rough’ substrata in backscatter maps. Because these data can be mapped at sea, targeted sampling at finer scales can be planned and implemented in ‘real-time’.

Such information allows interactions of fishers with fishery habitat to be understood at the level of primary biotopes, for example, the two types of fishing grounds that make up the Big Horseshoe Canyon SE terrace. The first of these is a mosaic of sediment and consolidated material that makes up most of the terrace that slopes gently between 300 and 600 m depth over a horizontal distance of about 9 km. This area, being clear of rough rocky ‘reefs’, provides good access for trawlers to catch a suite of upper slope species including pink ling. At the western margin of the terrace at the same depth range, the upper edge of a relatively steep slope forms the second ground type. This habitat descends to the base of the canyon at 850 m depth and is composed of patches of rough rocky bottom that emerges from surrounding sediments. It can be fished by static gears targeting a range of species, particularly pink ling, but does allow limited access to trawls. Although the extact boundaries of fishery habitats occurring at this level may be ill-defined - often representing transition zones between sediments and areas of rock reef - they provide the basis for estimating percentage areas of habitats at a scale relevant to spatial management planning. For example, management goals that specify target areas of habitat types to be contained within fishery closed areas or biodiversity conservation reserves.

3.3 Mesohabitat scales (10 m to 1 km)

Adding ‘ground-truth’ sampling to acoustic facies provides habitat resolution at the next level - Secondary Biotopes at Level 5. Ground-truthing includes observing the predominant elements of physical substrata and geomorphology and their fine-scale distribution using video and evaluating the composition of substrata from physical collections. Six Secondary Biotopes were identified at the Big Horseshoe Canyon SE terrace (Table 1). Sediments consisted of homogeneous calcareous muddy sands that form large unrippled patches to approximately 1 300 m in length at the shallower terrace sites and irregular (bioturbated) patches to approximately 900 m in length at the deeper sites. Rubble and debris of extensively burrowed claystones, mostly composed of gravel and pebble sized clasts, but some of cobble or boulder size, formed mosaics of numerous smaller patches to approximately 660 m in length. These were interspersed with sediments mainly around the southern perimeter of the terrace. Exposed sedimentary claystone rock on steep slopes at the western margin of the terrace forms relatively small patches (to 243 m in length) of subcrop and outcrop in distinct elongate horizontal ridges interspersed with patches of sediment and rubble or debris.

Level 5 is the minimum resolution level necessary for resolving habitat boundaries and patch structure for monitoring, and therefore mapping, during surveys because the high spatial variability encompassed at larger scales will obscure identification of impacts on habitat resulting from its use, as well as any benefits such as restoration resulting from management intervention. This is the basis for establishing animal-habitat associations at lower levels of habitat description and provides a resolution at which to understand the significance of habitat types, such as what defines ‘essential fish habitat’ (i.e. what limits populations in any way, sensu Steneck et al. 1997 and references therein). Optimizing the ‘ground-truth’ - targeted physical sampling of acoustic facies - is important for the execution of cost-effective surveys (Kloser et al. 2002).

Mesohabitat scale is also the size of seabed features that experienced fishers are familiar with and operate on. Their knowledge at this level is the basis for successfully targeting their fishing effort at features that result in aggregation of certain species in commercial concentrations (Bax and Williams 2001). Knowledge of habitat variability at this scale is therefore necessary for management areas to be defined without unnecessarily excluding fishers from important parts of larger fishing grounds. Sector-specific (gear-specific) fishery management intervention at this scale could correspond to clearly delineating claystone-based habitats at the western edge of the Big Horseshoe Canyon SE terrace.

3.4 Macrohabitat scales (10 m to 1 km)

Biological Facies, Level 6 in the scheme, at the macrohabitat scale are described by the conjunction of information on the dominant fauna with that on the physical seabed structure. Fifteen predominant biological facies were observed on the terrace. These included a sedentary fauna composed mostly of low (< 10 cm) encrusting sponges, anemones and sand-dwelling sponges that were the primary epifaunal inhabitants of unrippled muddy sands. Infaunal bioturbators - including ranellid gastropods and Latreillopsis petterdi - appeared to be abundant on the highly irregular (bioturbated) muddy sands. A mobile fauna including hermit crabs was also frequently observed. Small encrustors and erect epifauna were the most commonly observed biological facie associated with claystone debris or rubble. Beds of small sponges were also attached to this substratum where it was present on the steeply sloping seabed, and to debris or rubble composed of larger cobble and boulder sized clasts. The facies representing the greatest density of epifauna, the largest-sized individuals, and possibly the greatest biodiversity, were beds of small and large sponges associated with subcropping and outcropping claystone rock on the steeply sloping seabed.

Spatial management in the marine environment is ultimately directed at the biological inhabitants of habitats - to conserve biological diversity and local ecosystems or protect particular species (often commercial fishes for fishery management). Understanding animal-habitat associations will therefore require surveying at macrohabitat (and microhabitat) scales because these are the scales at which animal distributions vary, impacts can be recognised and quantified, and at which monitoring must occur.

3.5 Microhabitat scales (< 1 m)

Microhabitats, Level 7, represent the lowest level in the hierarchy. Those observed by video during the study are crevices, cracks edges and ledges associated with rocky outcrops and subcrops, irregular features such as pits and mounds associated with bioturbated sediments, and erect epifauna - mostly sponges - also associated with rocky outcrops and subcrops. The abundance of crevices, cracks, edges and ledges results from the combination of high seabed slope (to 15o) that exposes claystone, which is buried in sediment on flatter bottoms, and the pronounced up-slope dip, or tilt, in the rock that results in the down-slope rock faces being slightly elevated. These are the structured microhabitats with which high densities of pink ling were associated. Much of the claystone exists as detached flat boulders; those visible (not embedded in sediment) averaged 144 cm by 78 cm in size with the largest being 220 cm by 150 cm (n = 25).

Observing and understanding fishing impact must also occur at these scales. Video observations showed physical impacts occur when bottom trawls ‘hook-up’ on claysone boulders (or ‘slabs’) by turning and moving loose pieces. There is evidence of fishing impacting the habitat of the fish being targeted, that is at least partly irreversible. Understanding of vulnerability therefore relies on surveying at macrohabitat and microhabitat scales with extrapolation to primary biotope or geomorphic unit scales by mapping, or to provincial scales based on knowledge of regional geology (Bax and Williams 2001). The key attribute for understanding the impact on rocky claystone habitats is that these rock types are sedimentary and therefore friable, forming loose claystone boulders (‘slabs’), many of which are only partially embedded in sediments. They form large, although unquantified, fractions of mesohabitats and they, together with their attached epifauna, are movable or removable by trawls.

Targeted harvesting of aggregated pink ling populations on rocky habitat by static fishing methods is another form of impact and knowledge of such habitat associations may be necessary for meaningful stock assessment off southeastern Australia (Thompson in review). Off southern Africa, the combination of targeted trawl and demersal longlining resulted in severe depletion of kingklip (the closely related Genypterus capensis) (Punt and Japp 1994). While it is not known whether the different methods were targeting different habitats, Punt and Japp reported that reduced trawl catches were attributed by trawl operators to the systematic removal of the aggregated kingklip spawner stock by longline fishing. The relevance of information on multi-scale habitat distributions and species-associations to spatial management in the SEF is that at present (2003) the fishery is experiencing a large expansion in longline effort and an areal expansion of trap fishing to target pink ling at the same time trawl effort is expanding on the upper slope. The consequence is that all habitat types used by pink ling will be commercially fished and many of them with increasing effort.

Assessing the success of conservation measures requires repeated surveying to monitor changes in size and abundance of epifauna, and the distributions of mobile habitat features - particularly sediments. In deepwater, this must be done at microhabitat scales by photography in conjunction with near-seabed acoustic mapping.

4. FISHING INDUSTRY’S KNOWLEDGE OF HABITAT DISTRIBUTIONS

Habitat distributions at ‘intermediate’ scales - mega- and mesohabitat - are not known for the vast majority of the continental shelf and slope seabed around southeast Australia. Techniques for using surrogate variables to reliably predict the distributions of habitats and components of biodiversity at these scales are under active development (Kloser et al. 2001b) but substantial resources are required before scientific mapping at intermediate scales can be extrapolated over large areas.

However, these are the scales at which commercial fishers know the seabed and their working maps, typically in vessel electronic trackplotters, in the context of the hierarchical classification framework being used for the SER (Table 1), are a mix of Levels 3-6 in the habitat classification scheme. The utility of fishers’mapping data, if collected in the right form, is as it can possibly provide interpreted habitat information (distribution, boundaries, sizes, generalized geology and community types) at megahabitat scale or finer. As well, fishers collectively have near-complete coverage for the continental shelf and slope (from about 100 m out to about 1 300 m depth) at provincial scales.

A project between the CSIRO and the trawl and non-trawl sectors of the southeast Australian offshore fishing industry (Williams and Bax 2003b) was started in 2001 with the explicit aim of incorporating fishers’ knowledge of the seascape into strategic management planning. Industry executives supported the project primarily because they viewed it as a way to participate directly in the forthcoming, but then unspecified, spatial management process. It was argued that with their information systematically collected and rigorously evaluated fishers would be able to critically evaluate proposed spatial management plans, and push for management agencies to have clearly defined and measurable aims for their proposed management options. Equally importantly, these data provide industry with a synoptic view and a more detailed understanding of the habitats types that sustain the productivity of their fisheries. There is support to use these data to contribute to both the initial identification and subsequent selection of MPA sites. However, although involvement of industry data in this way has clear potential to enhance conservation, fishers remain uncertain about the consequences for them and therefore are uncertain about how, or indeed whether, to contribute their data.

5. CONCLUSIONS

We provide examples to illustrate a range of relevant spatial scales at which information on deep seabed habitats exists (Table 1) and suggest that it is the collective understanding of these scales - the seascape perspective - that enables specific management goals to be defined and their success to be evaluated. However, our examples are from one of the few surveys of deep shelf and slope habitats off Australia and it is unlikely that additional areas will be surveyed over the range of spatial scales needed in the timeframes (months to a few years) during which wide-ranging spatial management intervention is being planned for the Australian Marine Jurisdiction. Another prospective way of understanding habitat distributions at a regional scale would be through partnership with the offshore fishing industry. Fishers’ mapping data, if collected in the right form, could provide interpreted habitat information at useful spatial scales for the continental shelf and slope (from about 100 m out to about 1 300 m depth) with ‘provincial scale’ coverage. Including fishers’ knowledge in spatial management planning for a seascape best known to them is perhaps the best way to gain their acceptance and understanding of conservation objectives and for these to deliver fishery benefits through informed management of fishery habitat.

6. ACKNOWLEDGEMENTS

We gratefully acknowledge the research work of several colleagues in related fields that provides background for this paper, especially Vincent Lyne and Peter Last of CSIRO Marine Research. The habitat mapping methodology development survey that generated the example data used here was funded jointly by Australia’s National Oceans Office and CSIRO Marine Research. The CSIRO-industry mapping project was funded jointly by CSIRO Marine Research and the Fisheries Research and Development Corporation.

7. LITERATURE CITED

ANZECC TFMPA 1999. Understanding and applying the principles of comprehensiveness, adequacy and representativeness for the NRSMPA, Version 3.1. Report generated by the Action Team for the ANZECC Task Force on Marine Protected Areas. Marine Group, Environment Australia, Canberra.

BDAC 2001. Biodiversity Conservation Research: Australia’s Priorities. Environment Australia, Canberra. Australia and New Zealand Environment and Conservation Council and Biological Diversity Advisory Committee.

Bax, N.J. & A. Williams 2001. Seabed habitat on the southeast Australian continental shelf - context vulnerability and monitoring. Marine and Freshwater Research 52, 491-512.

Commonwealth of Australia. 1998. Australia’s Oceans Policy, vols. 1 and 2. Canberra: Environment of Australia.

CSIRO Marine Research 2001. Rapid assembly of ecological fish data (community composition and distribution) for the south-east marine region. Report by CSIRO Marine Research, Museum Victoria, Australian Museum and NSW Fisheries for the National Oceans Office, November 2001.

Exon, N. & P. Hill 1999. Seabed mapping using multibeam swath-mapping systems: an essential technology for mapping Australia’s margins. AGSO Journal of Australian Geology and Geophysics, 17:1-16.

Greene, H.G., M.M. Yoklavich, R.M. Starr, V.E. O’Connell, W.W. Wakefield, D.E. Sullivan, J.E. Jr McRea & G.M. Cailliet 1999. A classification scheme for deep seafloor habitats. Oceanologica Acta 22:663-678.

Kloser, R.K., A. Williams & A.J. Butler 2001a. Acoustic, biological and physical data seabed characterisation. Marine Biological and Resource Surveys of the South East Region, Progress Report 1 to the National Oceans Office.

Kloser, R.K., A. Williams & A.J. Butler 2001b. Acoustic, biological and physical data seabed characterisation. Marine Biological and Resource Surveys of the South East Region, Progress Report 2 to the National Oceans Office.

Kloser, R.K., G. Keith, T. Ryan, A. Williams & J. Penrose 2002. Seabed biotope characterisation in deepwater - initial evaluation of single and multi-beam acoustics. Proceedings of the Sixth European Conference on Underwater Acoustics, ECUA ‘2002, Gdansk, Poland 81-88.

Koslow, J.A., K. Gowlett-Holmes, J. Lowry, T. O’Hara, G. Poore & A. Williams 2001. The seamount benthic macrofauna off southern Tasmania: community structure and impacts of trawling. Marine Ecology Progress Series 213, 111-125.

Langton, R.W., P.J. Auster & D.C. Schneider 1995. A spatial and temporal perspective on research and management of groundfish in the northwest Atlantic. Reviews in Fisheries Science 3, 201-229.

Last, P., V. Lyne, G. Yearsley, D. Gledhill, M. Gomon, T. Rees and W. White 2005. Validation of the national demersal fish datasets for the regionalization of the Australian continental slope and outer shelf (> 40 m depth). Report for Australia’s National Oceans Office, 2005.

NOO 2003. The draft South-east Regional Marine Plan; Implementing Australia’s Oceans Policy in the South-east Marine Region. Australia’s National Oceans Office.

Punt, A.E. & D.W. Japp. 1994. Stock assessment of the kingklip (Genypterus capensis) off South Africa. South African Journal of Marine Science, 14:133-149.

Roff, J.C. & M.E. Taylor 2000. National frameworks for marine conservation - a hierarchical geophysical approach. Aquatic Conservation: Marine and Freshwater Ecosystems 10:209-223.

Simberloff, D. 1998. Flagships, umbrellas and keystones: is single-species management passé in the landscape era? Biological Conservation 83:2247-2257.

Steneck, R.S., R.W. Langton, F. Juanes, V. Gotceitas & P. Lawton 1997. Response: The interface between fisheries research and habitat management. North American Journal of Fisheries Management, 17:596-598.

Thompson, R. in review. Yield modelling for pink ling in the South East Fishery. Marine and Freshwater Research.

Ward, T. & E. Hegerl 2002. Implementing ecosystem based management for the Australian ‘Southern and Eastern Scalefish and Shark Fishery’. WWF Australia Report, 02/04, WWF Australia, GPO Box 528, Sydney, NSW, 2001.

Williams, A & N. Bax 2001. Delineating fish-habitat associations for spatially-based management: an example from the south-eastern Australian continental shelf. Marine and Freshwater Research 52:513-536.

Williams, A. & N. Bax 2003a. Integrating fishers’ knowledge with survey data to understand the structure, ecology and use of a seascape off southeastern Australia, pp.238-245. In Haggen, N., C. Brignall & L. Wood (Eds). Putting Fishers’ Knowledge to Work. UBC Fisheries Centre Research Report, vol. 11.

Williams, A. & N. Bax 2003b. Involving fishers’ data in identifying, selecting and designing MPAs: an illustration from Australia’s Southeast Region. Proceeding of the World Congress on Aquatic Protected Areas, Cairns, August 2002.

Williams, A, R.J. Kloser, and B.A. Barker 2004. Mapping, understanding and managing fishery habitat: a case study of the commercial pink ling (Genypterus blacodes: Ophidiidae) off SE Australia. Proceedings of the ICES Annual Science Conference, Vigo, Sept 2004

In situ observations of deep-water fishes in four canyons off the Georges Bank, NW Atlantic

F. Uiblein^[31], M. Youngbluth^[32], C. Jacoby^[33], F. Pagès^[34] M. Picheral^[35] And G. Gorsky^[35]

1. INTRODUCTION

The distribution and behaviour of fauna living in the area of highly structured continental slopes deserve special ecological interest as the physical conditions arising from interactions between topography and hydrology should impose considerable influences on the relationships among animals living close to the bottom and those in the adjacent pelagic region. Deep-sea canyons may stimulate bentho-pelagic interactions in particular through upwelling events, that transport larvae of bottom-dwelling fauna into the open water or enhance the horizontal transport and trapping of vertically migrating organisms on to the shelf (Tommasa et al. 2000). Further, downslope currents associated with the tidal cycle may transport nutrients to deeper waters where they may be used by benthic, benthopelagic and "pseudoceanic" fauna in the areas of submarine canyons. The ecological significance of bentho-pelagic interactions in the areas of steep, structured slopes with increased biomass, productivity and variation in local diversity has been emphasized in recent studies in the Canary Islands, Eastern Central Atlantic (e.g. Uiblein et al. 1996, 1998, Bordes et al. 1999, Uiblein & Bordes 1999, Ramos et al. 2001, Wienerroither 2003, 2005).

Behavioural studies of deep-sea macrofauna are rare and deserve more attention in the future as they can reveal important ecological relationships and distinct adaptations among individual species or populations. For example, recent in situ investigations of steep slopes and canyons in the Bay of Biscay, North-east Atlantic, have revealed significant inter- and intraspecific differences among seven demersal fish species in habitat use, activity level, locomotion behaviour and disturbance response (Uiblein et al. 2003).

This paper presents preliminary information on identification of midwater and demersal fishes to the lowest possible taxon, their spatial distribution and abundance patterns and their behaviour in four deep - water canyons. Data were collected with a manned submersible and an underwater video profiler. The basic study question was: "Does increased transport of organic material towards the bottom of deep submarine canyons enhance local aggregation and benthopelagic interactions of fishes?" Differences among canyons in sedimentation rates should be reflected in local fauna composition, distribution and behaviour.

2. METHODS

During the cruise in September 2002 a total of 24 submersible dives were conducted in four canyons off the Gulf of Maine (Figure 1; for further information see also: <http://www.at-sea.org/missions/maineevent3/synopsis.html>). During descent to the bottom (» 900 m depth) and ascent to the surface, the vertical distribution of mesopelagic fishes and other macrofauna was recorded. In six dives (Table 1), after arrival at the bottom, the immediate surrounding was explored for a short time period to record occurrence, abundance and behaviour of demersal fishes and epibenthic invertebrates. Fish identifications and behavioural activities were based on direct observation through the sphere of the submersible as well as video recordings.

The distribution patterns of particular distribution patterns of particular matter were obtained from vertical transects with an underwater video profiler (UVP, Figure 2). The UVP is a self-contained, battery-driven system that collects - among other data - optical information about the size, shape and abundance of objects larger than 60 m at a rate of 12-25 Hz (Gorsky et al. 2002).

3. RESULTS

A total of 21 fish taxa encountered in 10 dives (Table 1) were videotaped and their behaviour analysed (Tables 2, 3, 4). The most frequently observed pelagic taxa were myctophids followed by nemichthyids and serrivomerids, paralepidids, and the Atlantic eelpout (Melanostigma atlanticum). Myctophids, nemichthyids and Melanostigma atlanticum ranged vertically down to the benthic boundary layer and were in some instances encountered immediately above the bottom (Figure 3). Considerable differences in vertical distribution of these fishes occurred between day and night as well as among canyons (Figure 3).

The bristlemouth (Cyclothone sp.), which is supposed to be the most abundant fish taxon, could not always be readily identified because of its relatively small size and cryptic coloration. However, Cyclothone were observed frequently in the water column as well as close to the bottom. Also myctophids and macrozooplankton taxa including the siphonophore (Nanomia cara) and the ctenophore (Beroe sp.) were observed close to the bottom in some dives (Figure 3).

TABLE 1
Summary of canyon dives with the Johnson-Sea-Link submersible during which video recordings for behavioural studies of fishes were obtained

A - Atlantis Canyon, H - Hydrographer Canyon, L - Lydonia Canyon, O - Oceanographer Canyon; dives used for exploration of the bottom are emphasized in bold

TABLE 2
List of videotaped fish taxa ordered according to open water or bottom occurrence p/b - open water/bottom

For abbreviations of canyons see Table 1

TABLE 3
Distribution of demersal fish taxa among canyons

Taxa with intraspecific aggregation formation emphasized in bold

The fishes encountered in the water column showed different activity patterns and the more active ones, in particular myctophids, showed disturbance responses to the submersible. Several stomiids (e.g. Chauliodus sloani), nemichthyids and Atlantic eelpouts (Melanostigma atlanticum) remained completely inactive and the latter were often observed drifting motionless with the body moving completely passively in the open water. Specimens of the Atlantic eelpout could be collected easily using the suction system and samplers on the Johnson-Sea-Link submersible and were transferred to glass tanks for further observation in shipboard laboratories (Figure 4).

Demersal fish taxa showed clear differences in occurrence and formation of aggregations among canyons (Table 3). The highest diversity was encountered in Hydrographer canyon. Deep-sea scavengers consisting of two shoaling shark species (Centroscyllium fabricii, Apristurus sp.), aggregations of cutthroat eels (Synaphobranchus sp.) and solitary longfin hakes (Phycis chesteri) were observed in close proximity at distances of less than one metre to each other. Patchy distributions of Synaphobranchus sp. and Glyptocephalus cynoglossus were observed in Atlantis Canyon and Oceanographer Canyon (Figure 5) respectively. In Lydonia Canyon only three taxa and no intra- or interspecific aggregations of demersal fishes were observed.

Among the most frequent behaviour shown by the demersal fishes were active forward locomotion in sharks and cutthroat eels and a completely inactive "resting" behaviour observed in a shark, a chimaerid, and two hakes, longfin hake, Phycis chesteri (Figure 6), and Merluccius sp. (Table 4). Cutthroat eel specimens were active showing typical anguilliform forward locomotion or lateral drifting on or above the bottom. Similar observations had been made earlier during submersible dives in the NE Atlantic (Uiblein et al. 2002). Some individuals showed a disturbance response, most probably reflecting a reaction to the light and sound of, and in some cases also to increased turbulence produced by the submersible (see also Uiblein et al. 2003). During the dive in Atlantis Canyon, up to seven deep-sea eels could be observed simultaneously through the sphere of the submersible.

TABLE 4
Observed locomotion behaviour and response to submersible in demersal fish taxa

4. DISCUSSION AND CONCLUSIONS

Both mesopelagic and demersal fishes showed variation in species composition and spatial distribution within and among the four deep-water canyons. Differences in vertical zonation may be largely related to diurnal migration activities among the pelagic species. In the benthic boundary layer close associations were observed among pelagic and benthic fauna and aggregation formation in demersal fishes. The major factor inducing such distribution patterns may be allochthonous food input that should be rather high in this zone due to particle sedimentation both from the water column and the adjacent slopes (Keller & Shepard 1978, Noble & Butman 1989, Parmenter et al. 1983).

The formation of a scavenger assemblage close to the bottom in Hydrographer Canyon may be closely associated with increased food abundance as suggested by the high turbidity (oungbluth et al. 2003). The impression of high scavenging activity in Hydrographer Canyon is also enhanced by the discovery of a yet unidentified pelagic shrimp in the open water a few metres above the bottom that carried a dead Cyclothone (Figure 7). Also in the Oceanographer Canyon particle concentration and fish density close to the bottom was relatively high. No aggregations occurred in Lydonia Canyon and there was a low demersal fish diversity. This result is consistent with a less active canyon with regard to particle concentration.

The inactive behaviour observed in several demersal fish species may reflect sit-and-wait foraging, passive predator avoidance or metabolic relaxation strategies that should be particularly successful on highly structured bottoms of deep-sea canyons. For instance, a dense aggregation of orange roughy (Hoplostethus atlanticus, Trachichthyidae) was recently discovered residing mostly inactively on the bottom of a slope canyon in the Bay of Biscay, NE Atlantic (Lorance et al. 2003). Apart from serving as foraging habitat or a refuge (oklavich et al. 2000), canyon bottoms may also be used by demersal fishes for spawning (Uiblein et al., 1996, 1998, Murdoch et al. 1990) or egg-brooding (Drazen et al. 2003).

In conclusion, deep submarine canyons may play a biologically important "interactive role" as a source and sink habitat for the surrounding shelf, slope and pelagic areas. Future studies should investigate the overall faunal composition and the spatial and trophic interactions in the boundary layer in more detail using submersibles, moored cameras, acoustic surveys, and sediment traps among other methods. Consideration should also be increasingly given to long-term ecological monitoring and sustainable management of marine resources in the area of deep-sea canyons.

5. ACKNOWLEDGEMENTS

The Gulf of Maine cruise was made possible through a grant from the Biological Oceanography Program of the National Science Foundation with additional support from Harbour Branch Oceanographic Institution to Marsh oungbluth. Franz Uiblein received travel support by the Österreichische Forschungsgemeinschaft, projects 06/7060 and 06/7538, the University of Salzburg, Austria, and the Institute of Marine Research, Bergen, Norway.

6. LITERATURE CITED

Bordes, F., F. Uiblein, R. Castillo, A. Barrera, J.J. Castro, J. Coca, J. Gomez, K. Hansen, V. Hernandez, N. Merrett, M. Miya, T. Moreno, F. Perez, A. Ramos, T. Sutton & M. Yamaguchi 1999. Epi- and mesopelagic fishes, acoustic data, and SST images collected off Lanzarote, Fuerteventura, and Gran Canaria, Canary Islands, during cruise "La Bocaina 04-97". Inf. Téc. ICCM. 5: 1-45.

Drazen, J.C. 2003. Aggregations of egg-brooding deep-sea fish and cephalopods on the Gorda Escarpment: a reproductive hot spot. Biol.Bull. 205: 1-7.

Gorsky, G., L. Prieur, I. Taupier-Letage, L. Stemmann & M. Picheral 2002. Large particulate matter in the Western Mediterranean. I. LPM distribution related to mesoscale hydrodynamics J. Mar. Syst. 33-34: 289-311.

Keller G.H.& F.P. Shepard 1978. Currents and Sedimentary Processes in Submarine Canyons off the Northeast United States. Sedimentation in Submarine Canyons, Fans, and Trenches.15-32.

Lorance, P., F. Uiblein & D. Latrouite 2002. Habitat, behaviour and colour patterns of orange roughy Hoplostethus atlanticus (Pisces: Trachichthyidae in the Bay of Biscay. J.Mar. Biol. Assoc.UK 82: 321-331.

Murdoch, R., R. Guo & A. McCrone 1990. Distribution of hoki (Macruronus novaezelandiae) eggs and larvae in relation to hydrography in eastern Cook Strait, September 1987. NZ.J.Mar.Freshw.Res. 24: 529-539

Noble, M. & B. Butman 1989. The structure of subtidal currents within and around Lydonia Canyon: Evidence for enhanced cross-shelf fluctuations over the mouth of the canyon. J. Geophy Res. 94:8091-8110.

Parmenter, C.M., M.H. Bothner, & B. Butman 1983. Comparison of four sediment-types deployed in Lydonia Canyon. EOS 64:1052

Ramos, A.G., F. Bordes, T. Moreno, F. Uiblein, J. Carrillo, A. Barrera, R. Varó Domenech, R.A. Bravo Castro, Rodríguez E. Avendaño, J. Coca, A. Mesones, U. Gancedo & R. Wienerroither 2001. Evaluación acústica de los recursos epipelágicos en aguas de la Plataforma insular del archipelago canario. Resultados de la Campaña ECOS 0499. Viceconsejería de Pesca. Consejeria de Agricultura, Pesca y Alimentación, Las Viceconsejería de Pesca. Consejeria de Agricultura, Pesca y Alimentación, Palmas, Gran Canaria, 73 pp.

Tommasa D.L., G. Belmonte, A. Palanques, P. Puig & F. Boero 2000. Resting stages in a submarine canyon: a component of shallow-deep-sea coupling? Hydrobiologia 440: 249-260.

Uiblein, F., F. Bordes & R. Castillo 1996. Diversity, abundance and depth distribution of demersal deep-water fishes off Lanzarote and Fuerteventura, Canary Islands. J.Fish Biol. 49: 75-90.

Uiblein, F., F. Bordes, R. Castillo & A. Ramos 1998. Spatial distribution of shelf- and slope-dwelling fishes collected by bottom longline off Lanzarote and Fuerteventura, Canary Islands. Mar.Ecol. 19: 53-66.

Uiblein, F. & F. Bordes 1999. Complex trophic interactions around ocean islands. Ocean Challenge, 9: 15-16.

Uiblein, F., P. Lorance & D. Latrouite 2002. Variation in locomotion behaviour in northern cutthroat eel (Synaphobranchus kaupi) on the Bay of Biscay continental slope. Deep-Sea Res. 49: 1789-1603.

Uiblein, F., P. Lorance & D. Latrouite 2003. Behaviour and habitat utilisation of seven demersal fish species on the Bay of Biscay continental slope, NE Atlantic. Mar.Ecol.Prog.Ser. 257: 223-232.

Yoklavich, M.M., H.G. Greene, G.M. Cailliet, D.E. Sullivan, R.N. Lea & M.S. Love 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fish.Bull. 98: 625-641.

Youngbluth, M., G. Gorsky & M. Picheral 2003. Coupling between macrozooplankton and particulate material within four submarine canyons that indent the seaward shelf of Georges Bank. Poster presentation, Conference in Gijon, Spain The Role of Zooplankton in Global Ecosystem Dynamics: Comparative Studies from the World Oceans, May 20- 23, 2003.

Wienerroither, R. 2003. Species composition of mesopelagic fishes in the area of the Canary Islands, Eastern Central Atlantic. Inf. Tec.ICCM 9: 1-110.

Wienerroither, R. 2005. Meso- and bathypelagic fishes of the Canary Islands: an annotated species list, species composition, and biogeographic distribution. PhD thesis, Paris Lodron University of Salzburg, Faculty of Natural Sciences, 141 pp.