Sedimentary facies, geomorphic features and habitat distribution at the Hudson Canyon head from AUV multibeam data
.
Martina Pierdomenicoaa, Vincent G. Guidabb, Leonardo Macellonicc, Francesco L. Chiocciaa, Peter A. Ronadd, Mary I. Scrantonee, Vernon Asperccand Arne Dierckscc
a) aDepartment of Earth Sciences, Sapienza University of Rome, Piazzale Aldo Moro 5, 00185 Rome, Italy.
b) bNOAA NE Fisheries Science Center, 74 Magruder Road, Highlands, NJ 07732, United States.
c) cNational Institute for Undersea Science and Technology, University of Mississippi, 310 Lester Hall, University (MS) 38677, United States
d) dDeceased, formerly Institute of Marine and Coastal Sciences, Rutgers University, 71 Dudley Road, New Brunswick, NJ 08901-8521, United States
e) eSchool of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY 11794, United States
Corresponding author: Martina . +39 06 4991 4935
ABSTRACTbstract
Mapping of physical benthic habitats at thehead of Hudson Canyon was performed by means of integrated analysis of acoustic data, video surveys and seafloor sampling. Acoustic mapping, performed using AUV-mounted multibeam sonar, provided ultra-high resolution bathymetric and backscatter imagery for the identification of geomorphological features and the characterization of surficial sediments.Habitatcharacterization in terms of seafloor texture and identification of benthic and demersal communities was accomplished by visual analysis ofstill photographs from underwater vehicles. Habitat classes were defined on the basis of the seafloortexture observed on photosand then compared with the geophysical data in order to associate habitats to acoustic classes and/or geomorphological features. This enabled us to infer habitat distributionon the basis of morpho-acoustic classes and extrapolate results over larger areas.Results from bottomtrawling were used to determine the overall biodiversity within the identified habitats.Our analysis revealed a variety of topographic and sedimentologicalstructures that provide a wide range of physical habitats. A variety of sandy and muddy substrates, gravel patchesand mudstone outcrops host rich and varied faunal assemblages, including cold-water corals and sponge communities. Pockmark fields below 300 m depth suggest that methane-based chemosynthetic carbonate deposition may contributes to creation of specific benthic habitats. Hummocky terrain hasave been delineated along the canyon rims and associated with extensive, long-term burrowing activity by golden tilefish (Lopholatiluschamaeleonticeps).These resultsshow the relationships of physical features to benthic habitat variationariability, support the notion of the area as a biodiversity hotspot and defineessential habitats for planning of sustainable regional fisheries.
Keywords:
Hudson submarine Canyon, seafloor mapping, benthic habitat, backscatter imagery.
Introduction
1. Introduction
The Iincreasing human impacts on the marine environment raise the need for to promote management approaches capable of sustainingthe health of ecosystems through the preservationpreservationof their structure, functioning and key processes (De Young et al., 2008). On the other hand, any sustainable approach requires a good knowledge of the spatial distribution and ecological functioning of marine ecosystems, over a range of different scales (Cogan et al., 2009; TallisandPolasky, 2009; Cogan et al., 2009; Salomidi et al., 2012). This is particularly true for the deep sea, whose ecosystems are still poorly known (Wilson et al., 2007; Harris and Baker, 2011; Heymanand Wright, 2011).
In this context,the mapping of benthic habitats has become a major tool in the assessment and monitoring of marine ecosystemsand planning of Marine Protected Areas (Pickrill and Todd, 2003;Harris et al., 2008;Muñoz et al., 2009; Howell et al., 2010; Copeland et al., 2013).The process of producing seafloor habitat maps requires the integration of marine biology, geology, oceanography and geophysics, in order to produce simplified representations of the seafloor related to the distribution of biological communities at different spatial scales. Thetermnotion of‘habitat,’traditionally related to the place where an organism ordinarily can be found, has evolved in the direction of a spatially recognizable area characterized by physical and environmental conditions thatsupport a particular biological community, together with the communityitself (Valentine et al., 2005; Coggan et al., 2007; Valentine et al., 2005). The rapid development in recent years of high-resolution seafloor mapping techniques, such as Side Scan Sonar (SSS) and MultibeamEchosounders (MBES), significantly significantly increased theboosted our capability to image and map the seafloor over large areas (Brown andBlondel, 2009; Brown et al., 2011a). Full coverage bathymetry and the derived information of the seafloor integrated with ground truth data enableallow the recognition of different habitats and provide an interpretative tool for predicting their distribution and those of the marine resources that they support (Brown et al., 2011b and references therein). The recent use of cutting-edge technologies such as Remotely Operated Vehicles (ROV) and Autonomous Underwater Vehicles (AUV) as platforms for acoustic systems led to a significant improvement in the scale to which deep- sea habitats can be identified and in the description of the associated biota (Grasmueck et al., 2006; Dolan et al., 2008; Vertino et al., 2010; Huvenne et al., 2011; Macelloni et al., 2013).
Substratum type, topographic relief, sediment composition, and geomorphology of the seabed arehave been identified as important descriptors of biological patterns (Buhl-Mortensen et al., 2009;Harris and Baker, 2011).Seafloor complexity and habitat heterogeneity are recognized to play a key role in enhancing the faunal biomass and biodiversity in deep- sea environments (Buhl-Mortensen et al., 2010;Vanreusel et al., 2010;De Leo et al., 2014). Characterization of the seabed in terms of terrain parameters such as slope, aspect or curvature and geomorphic features, along with the segmentation of MBES and SSS data into acoustic facies (i.e., regions showing similar acoustic properties or features), commonly referred to as Acoustic Seabed Classification (ASC, Anderson et al., 2008), thus offers a valuable tool for delineating regions of the seafloor that may support specific communities and thus provide distinct habitats (Wilson et al., 2007;Buhl-Mortensen et al., 2009;SaviniandCorselli, 2010).
We present a case study ofbenthic habitat mapping in the upper reach of Hudson Canyon based on the integrated analysis of acoustic and groundtruth data.
Hudson Canyon, located about 180 km SE of New York City, is under evaluation for the assignment of HAPC (Habitat Area of Particular Concern) status and represents a fisheries and biodiversity hot spot (Stevenson et al., 2004; Mayo et al., 2009). It is also the focus of a collaboration between the NOAA Northeast Fisheries Science Center, the Mississippi Mineral Research Institute (MMRI), the National Institute for Undersea Science and Technology (NIUST), Stony Brook and Rutgers Universities. This collaboration aims at creating an integrated database that includes existing and newly collected data, such as acoustic mapping, visual ground-truthing, hydrographic, sedimentological and trawl data collections, as a basis for the study of benthic habitats and for the development of habitat suitability models for fisheries species.
The aims of this study are therefore:
(1) To produce the first detailedmap(tens of meters) of the benthic habitats at the Hudson Canyon head using very high-resolution acoustic data from an AUV-mounted MBES;
(2) To derive a quantitative relationship between acoustic parameters and ground truth results to classify the study area into categories related to different seafloor characteristics;, and
(3) To qualitativelyoutline the biodiversity of the area by using information fromphotos and video imagery andtrawl samples.s.
2. Hudson Canyon
Hudson Canyon (Fig.1)is the largest submarine canyon on the eastern U.S. Atlantic margin and one of the largest inof the world (Ericson et al., 1951; Pratt, 1967). It extends for over 400 km, from the outer shelf down to the upper continental rise at about 3500 m depth (Heezen et al., 1959; Pratt, 1965). The canyon head deeply incises the continental shelf starting at 80 m water depth, about 40 km shoreward of the shelf margin, and is composed of two branches, NW-SE and N-S oriented, that merge at a depth of about 120 m into a segment oriented parallel to the main canyon course (Stanley and Freeland, 1978; Butman et al., 2006).
From the head to the base of the continental slope (about 2200 m depth), the canyon is up to 12 km wide and up toan 1100 m incised depth with respect to the surroundings (Butman et al., 2006). The canyon displays a flat thalweg,500 to 900 m wide.The canyon walls are characterized by multiple ridges perpendicular to the thalweg axis, separated by a dense network of gullies (Twichell and Roberts, 1982; Butmanet al., 2006).
The Hudson Shelf Valley, a shallow trough extending across the continental shelf, connected the Hudson River to the canyon head during glacial low stands (Knebel et al., 1979).Holocene gravel and coarse sand-depositsof fluvial originare present at the canyon head and show evidence ofreworking by currents and bioturbation(Schleeand Pratt, 1970). These coarse relict sedimentsoccur down to a water depth of 130 to 175 m, where a sharp boundary exists with the present muddy cover. This appears to record a long-term separation of different energy zones, i.e. below theboundary current speeds necessary to erode fine sediments decrease significantly in frequency and intensity (Stanley and Freeland, 1978).
The general pattern of circulation in the area involvesshelf watersand the warmer and more saline slope-waters,separated by the abrupt gradient called the shelf-slope front (Wright, 1983; Chapman, 1986). On average, shelf watersmove towards the southwest parallel to bathymetric contoursat speeds of 5-10 cm/s at the surface and 2 cm/s or less at the bottom (BurrageandGarvine, 1988; Stevenson et al., 2004). Stratification of the water column occurs over the shelf and within the top layer of slope water during the spring-summer and persistsuntil autumn, while a permanent thermocline exists in slope waters from 200 to 600 m depth (Aikman, 1984).Moreover, the “cold pool,”,i.e. a bottom shelf water characterized by a minimum temperature of 1.1 to 4.7°C (Amstrong, 1998), is present on the continental shelf bottom at depths ranging between 40 and 100 m, from the spring to early fall.The upper reach of the canyon is affected by complex oceanographic dynamics including internal wavesand bottom currents flowing parallel to the canyon axis (Keller et al., 1973; Hotchkiss andWunsch, 1982).Measurements of bottom currentsvelocities within the canyon revealed a semidiurnal reversal of flow (Keller et al., 1973),suggestingactiveresuspension andtransport of fine material through the canyon to the outer continental rise (Keller and Shepard, 1978; Shepard et al., 1979).
At the head of the canyon, intense mix due to breaking and dissipation ofinternal waves was observed by Hotchkiss and Wunsch (1982). Moreover, Church et al. (1984) suggested that the interaction of the canyon with the Mid-Atlantic Bight (MAB) ‘cold pool’ water may promote enhanced nutrient exchange and biological production. Climatological CZCS satellite observations of surface chlorophyll indicate enhanced surface primary productivity at the regions near Hudson Canyon and other shelf break canyons (Ryan et al., 1999).
As observed for other canyons of the northwestern Atlantic margin (Hecker et al., 1983), Hudson Canyon is suitable to host a rich and more varied fauna compared to the surrounding shelf and slope areas.Rowe et al. (1982) found that macrofaunal composition inside the canyon did not differ substantially from the adjacent slope (except for high densities within the canyon’s head). Nevertheless, commercial and recreational catches in the shelf areas surrounding Hudson Canyon indicate the occurrence of a great variety of demersal fishes and invertebrates(Jacobson et al., 2009; Mayo et al., 2009; Jacobson et al., 2009). Stevenson et al. (2004) reported intense bottom otter trawl activity for the period between 1995 and 2001 in the shelf areas around the canyon. Limited observations also suggest that the canyon increased concentrations of krill that attract larger numbers of marine mammals in the Hudson Canyon area (Greene et al., 1988).Hecker and Blechschmidt (1980) reported eEvidence of cold-water corals within Hudson Canyon. are reported by Hecker and Blechschmidt (1980), whoThey found abundant populations of the soft coral,Eunephthyafruticosa, in the deeper portion of the canyon. Solitary stony cold-water corals were observed on the shelf around Hudson Canyon and in the head of the Canyon (Packer et al., 2007).
Moreover, Wwide shelf areas around the head of the canyon display a unique rough topography with relief of 1-10 m.This irregular hummocky topography is attributed to seafloor erosion and burrowing activity of golden tilefish (Lopholatiluschamaeleonticeps) and associated species of crustaceans (Able et al., 1982; Twichell et al., 1985).
.
3. Data and Methods
To produce the benthic habitat map of the upper reach of Hudson Canyon we used a large varietyvariety of data collected between 2004 and 2011 in the framework of different projects. The primary datasetconsists of bathymetric and backscatter datafrom a Kongsberg EM2000 (200 KHz)multibeam sonar, mounted on a NIUST “Eagle Ray” Autonomous Underwater Vehicle (AUV). Groundtruth data include still photos and grab samples collected by the United States Geological Survey (USGS) Sea Bottom Observation and Sampling System (SEABOSS) towed video vehiclefor areas shallower than 200 m,and photos acquired by the NIUST “MolaMola” AUV at greater depths (Fig.2).In addition, demersal fishes and benthic megafaunal catches fromtrawlsprovided individuals for the taxonomical identification of the benthic fauna observed on photos and were used, along with the still photos, to define the overall biodiversity within the differentidentified habitats.All data used in this study are shown on the map in Fig.2.
3.1. Visual and sediment groundtruth data
To characterize andclassify the different types of habitat across the study area, we consideredtook into consideration photos and seafloor sediment samples collected during two cruises in 2004 and 2011.The 2004 dataset was collected aboard NOAA Ship Delaware II using the USGS SEABOSS towed video vehicle. SEABOSS has two video cameras (forward and downward looking), a downward looking 35 mm camera, and a modified Van Veen grab sampler. Quartz halogen lights provide illumination for the video, and an electronic flash unit provides lighting for still photography. Dual lasers provide accurate photographic range and scale information. The system is tethered and essentially “flown" over the seafloor by a shipboard operator while the support vessel is drifting. Images from both video cameras were recorded on tape, but were also viewed in real time, allowing collection of representative still photographic images. Each deployment of SEABOSS consisted of drift transects of continuous video of variable duration, but averaging 26 minutes apiece. Still photos of features of biological and geological interest were taken at irregular intervals as they appeared on video, with an average rate of one photo per minute.As SEABOSS hangs nearly vertically from the ship during deployment, ship-mounted GPS provided georeferencing for these photos. Stations for this cruise have beenchosen on the basis of previous low-resolution acoustic data in order to include canyon margins and walls, the thalweg and the adjacent shelf.In this study we only analyzed the still photos acquired by SEABOSS. A total of 727 photosfrom 16 video-transectswithin the study area (Fig.2) were analyzed. Sediment samples were taken by the SEABOSS Van Veen grab at the end of each transect and used for grain size analyseis, which. Grain size analyses for Seaboss samples were performed at the USGS Woods Hole Science Center according to standard methodologies practiced by that laboratory at the time of their collection (PoppeandPoloni, 2000).
The 2011 dataset was collected aboard NOAA Ship Henry B. Bigelow by using the NIUST “MolaMola” AUV. Two long baseline (LBL) and one ultra-short baseline (USBL) bottom moorings were placed on the canyon bottom to insure positional precision.MolaMola was programmed to travel at an altitude of 3.0 ± 0.2 m above the bottom and to take a still photo every 4 seconds.Four deployments of MolaMola within the study area (Fig.2) collected a total of 1638 photos. Stations were planned to include some physical habitats occurring in the deepest part of the study area that were not surveyed during previous cruises.
Habitat characterization was accomplished through the analyses of the still photos, that included enumeration of demersal fishes and macrofaunal invertebrates and description ofseafloor texture. In this study theseafloor textureis the main factor taken into account to classify habitats,following the approach suggested byGreen et al. (1999)for meso- and macro-habitats (i.e.,tens to few hundreds of meters). We thusdefined seven classes of benthic habitats, corresponding to as manyuchseafloor types identified on the photos: Gravel and cobbles;, Mudstone outcrops;,Sand with gravel;, Sand;, Muddy sand/Mud;, Sub-outcropping and outcropping rock; and Hummocky terrain (i.e., seafloor areas characterized by a rough micro-topography produced by heavily burrowed semilithified clay outcrops, overlainied by a veneer of fine sediment). To confirm and detail the nature of the seafloor corresponding to the different habitat classes,we used the sediment samples taken by the SEABOSS Van Veen grab at the end of each transect.An additional habitat class was defined for the photo-transects collected in correspondence of pockmark areas, where high concentrations of dissolved methane were measured in near bottom-water samples (Rona et al., 2009).Habitat classes weremapped along each transect and then compared with bathymetric and backscatter data.
.
3.2. Multibeam and backscattering data
Acoustic mapping of Hudson Canyon was accomplished during three cruises in 2007, 2008 and 2009, aboard the NOAA vessels Ronald H. Brown and Henry B. Bigelow. Ten deployments of “Eagle Ray” lastingof about 18 hours each allowed the mapping of an area of about 140 km2 between 85 and 700 m depth, corresponding to the first 25 km of the upper reach of the canyon (Fig.2). The vehicle traveled at a constant altitude of about 60 m above the seafloor following 150-m spaced parallel track lines aligned along the canyon axis. A single bottom mooring, kept approximately in the middle of survey area with disposable weights, provided a fixed georeference. A hydrophone array was used for communication between the ship and the AUV.
Raw acoustic data were processed with the Caris HIPS& SIPS® 7.0 software, allowing the correction of bathymetric data by accounting for sound velocity variations, tides, out-of-sequence beams and spikes. The filtered data were used to produce a digital elevation model (DEM) of the study area with a 3 m horizontal resolution (Fig.3A). A slope map (Fig.3B) was extracted from the bathymetric data and used to identify gradients in elevation that are indicative of specific topographic features (e.g., outcrops, pockmarks).Backscatter data were processed using Geocoder®, a software tool developed by Fonseca and Calder (2005). Raw backscatter data were radiometrically corrected to remove variable acquisition gains, power levels, insonification area and grazing angles; geometric corrections were applied to compensate for the navigation and transducer attitude and a feathering algorithm was used to reduce the seam artifact between overlapping lines during the mosaicking of the data. Final product was a “normalized” backscatter grid with a 1 m horizontal resolution (Fig. 3C).