Improving the quantitative estimation of trawling impacts from sidescan-sonar and underwater-video imagery
Hellenic Centre for Marine Research, PO Box 2214, 71003 Heraklion, Crete, Greece
Correspondence to C. J. Smith: tel: +30 2810 337752; fax: +30 2810 337822; e-mail: csmith{at}her.hcmr.gr
Smith, C. J., Banks, A. C., and Papadopoulou, K.-N. 2007. Improving the quantitative estimation of trawling impacts from sidescan-sonar and underwater-video imagery. – ICES Journal of Marine Science, 64: 1692–1701.The techniques of sidescan sonar and towed, underwater-video sled were assessed as rapid-assessment methodologies for investigating trawl impacts on the substratum. Sidescan sonar is able to image a swathe of
200 m with a resolution of 20 cm at a speed of 2–3 knots, and marks of trawl doors could be observed. The towed video system imaged a swathe of 1–2 m with a resolution of 1–2 cm at a speed of 1 knot, and trawl-door marks, scrape marks, local bioturbation features, and fauna could be observed. Multiple tows using both methodologies were carried out in two areas in Heraklion Bay, Crete. One area, experimentally trawled, was 80–90 m deep and characterized by mixed, maerly sediments; the other was a commercial trawl lane 200 m deep characterized by silty-clay sediment. Descriptions of the types of trawling feature and impacts caused by trawling were made for both areas. Images were analysed from the commercial deeper trawling ground for area assessment. For sidescan-sonar records, direction of trawling and trawl-mark density by category were estimated at periodic intervals along the track. For video, categories for trawl-mark density and level of bioturbation were estimated, along with the density of the crinoid Leptometra phalangium. Using geo-referenced positioning for each data point, area maps were constructed for each of the parameters, and correlations were tested between the different datasets. The use of the assessment techniques (characteristics, data usage, mapping, complementarity) in relation to trawling-impact studies is discussed, as well as possibilities for the use of the resulting data for management.
Keywords: Aegean sea, optical and acoustic imaging, sidescan sonar, towed video sledge, trawling impacts
Received 3 January 2007; accepted 26 June 2007.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
The impacts of trawling worldwide have been studied in detail for at least 15 years, and they include investigations into impacts on target and non-target species, on the physical and chemical environment of the seabed, and on benthic fauna (Hall, 1999; Kaiser et al., 2002; Hiddink et al., 2007). Studies have included, inter alia, monitoring studies, experimental impact studies and mesocosm disturbance studies, and results have generally shown that trawling causes a reduction in faunal structural diversity and abundance (Dayton et al., 1995; Jennings and Kaiser, 1998; Collie et al., 2000; Gislason et al., 2000; Kaiser et al., 2006). Even with detailed studies of the impacts of trawling on coastal and shelf ecosystems, there remains a need to develop quantitative and rapid methodologies for estimating the impact. Useful means of analysing and categorizing the data are also needed for coastal and fisheries management and conservation decisions.
Trawling alters the physical environment of the benthos by creating furrows or scars from trawl doors, scouring and flattening the seabed with the ground rope and weights, and redistributing or removing sediment and other material (Churchill, 1989; Riemann and Hoffmann, 1991; Schwinghamer et al., 1996). It is these physical changes that are visible to imaging systems such as sidescan sonar and underwater video. In sidescan-sonar images, and some multibeam-sonar images (high frequency), trawl marks are seen as tracks (door marks) across the seabed (Service and Magorrian, 1997; Schwinghamer et al., 1998; DeAlteris et al., 1999; Friedlander et al., 1999; Humborstad et al., 2004; Malik and Mayer, 2007). On video, more details of the plough marks, scrape marks, sediment heaps, and topographical flattening can be seen and easily differentiated from naturally roughened (bioturbated) untrawled areas (Service and Magorrian, 1997; Schwinghamer et al., 1998; Smith et al., 2000; Humborstad et al., 2004). Typically, sidescan sonar can cover 1 km2 h–1 with a resolution of 20 cm, and video 0.002 km2 h–1 with a resolution of 1 cm (Kenny et al., 2003; Smith and Rumohr, 2005), so large areas may be covered over time. Friedlander et al. (1999) showed that sidescan sonar can be used to assess trawling impacts over a wide area. Further, Pinn et al. (1998) and Humborstad et al. (2004) demonstrated how a combination of complementary methodologies is useful in assessing benthic communities and trawling impacts. Imaging technologies are now widely available within the scientific community and, with good sampling design and a little effort in sampling and analysis, the information they deliver can be presented in area maps and contour plots that may be useful for management. Friedlander et al. (1999) made a good first effort at mapping trawl-mark density over large spatial scales using sidescan sonar. The work we present here represents an attempt to improve on the basic methodology of the estimation of trawl density from sidescan, and also shows how to use a similar methodology for underwater video for rapid, independent methodologies for creating accurate broad-coverage maps of relative trawling intensity or impact, or of both parameters.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Study sites
Our study area was in Heraklion Bay, Crete, the principal area being south of Dia Island inside the bay (Figure 1). There, fishers follow a number of distinct commercial trawling lanes, one of which follows the 200 m isobath and narrows in a shallow valley behind Dia Island, where trawlers generally haul their nets. The lane is almost entirely defined by the bathymetry, and easily definable control sites exist near it. The sediments are characterized by relatively soft silty clays (median grain size 0.016–0.019 mm, with sand, silt, and clay at
9%, 88%, and 3%, respectively; Smith et al., 2000).
|
The second study area, off Gouves (Figure 1), is an example of shallower trawling ground (it is some 80–90 m deep). It is situated adjacent to other commercial trawling lanes within an area protected by rocky and calcareous algal reefs. Two study sites were investigated in the neighbourhood of Gouves, an experimentally trawled site, and an adjacent protected control site. Sediment composition was mixed, but was generally coarse sand with some mud in localized areas, with calcareous sand/rock fragments on the sediment surface (median grain size 0.10–0.13 mm, with sand, silt, and clay at
68%, 30.5%, and 2.5%, respectively; Smith et al., 2003). For this analysis, the Gouves site was used only for general observation of trawling impacts on coarser sediments.
Equipment and data acquisition
The sidescan-sonar system used comprised a Geoacoustics 196D Towfish, winch-mounted coaxial cable SS941 Transceiver, and a GeoPro LC Processor (all Geoacoustics, UK) connected to a DGPS system (Garmin GPS, Fugro DGPS signal). Weighted rope (30 m, weighing 30 kg) was attached to the cable 20 m in front of the towfish to act as a cable depressor. All sidescan surveys were carried out at a frequency of 410 kHz, with 98 m range on each side, i.e. a swathe of 196 m, giving a resolution of 15 cm and coverage of 1.0 km2 h–1.
The sidescan was deployed over the stern of the vessel, and a counting block was used to determine the amount of wire paid out. The position of the towfish was geo-referenced through continuous adjustment of the layback into the processing system, and adjusted as the tow height, water depth, or length of cable out changed. This was necessary because the towfish was mainly some 500–600 m behind the support vessel. Layback was calculated on a spreadsheet using the formula:
|
| (1) |
where L is the layback, C the length of cable out, D the depth, H the towfish height, and G the distance from the GPS antenna to the cable counter, with all lengths in metres.
Track recording began when the towfish was within some 20 m of the seabed and stopped when it left the seabed at the end of the run. Normal operating height was 8–15 m above the seabed, and towing speed was 2–3 knots. In the Dia Island area, 15 tows were carried out as curved transects crossing the fishing ground, or as transects along the axis of trawling both in and outside the trawling lane. Tows were carried out during both open and closed trawling seasons, compiling the data together in the knowledge that trawl marks are persistent (Smith et al., 2000). The exact direction of tow was to some extent dependent on the weather and the direction tows could be undertaken.
The video sledge used was the Fisheries Research Services (Aberdeen) design (Shand and Priestley, 1999). The camera was an Osprey (OE1360 Osprey Electronics, Aberdeen) low-light sensitive, colour camera, and was mounted on the sledge pointing obliquely forward, with two wide-angle 500 W underwater lighting units (Versabeam, Deep Sea Power & Light, Aberdeen). The view had previously been calibrated with a grid placed in front of the sledge so that the view was 65 cm wide at the bottom of the frame and 2 m at the top in oblique view.
Maximum resolution was 1–2 cm and area coverage 0.002 km2 h–1. The sledge was towed from the stern of the research vessel on a 12 mm trawl warp to which the electrical cables of the camera and lights were attached by quick-release ties. Floatation was added to the warp at the sled end of the cable to help keep the towing cable from disturbing the sediment in front of the sled. Tow speed was held steady at 1 knot, because at faster speeds the image was blurred and at slower speeds there was a tendency for the towing warp and umbilical to drop onto the seabed. In all, 10 video tracks were run in the Dia area, the sledge generally towed in a zig-zag pattern across the fishing lane from the unfished areas to its north and south.
The position of the towing vessel was recorded every 5 min (DGPS position), the video sledge was deployed on the seabed, and the output from the TV camera was recorded on videotape (S-VHS), along with a time signal from when the sledge touched the seabed to when it was retrieved. Layback could not be calculated accurately for the video, but a rough estimate was 200 m. Unlike the sidescan towfish, the video sled was often slightly off to one side of the vessel. Major features viewed were continually logged by the operators. Detailed analysis of the videotapes was carried out ashore.
Detecting, quantifying, and mapping trawling impacts at Dia Island
At playback in the laboratory, each sidescan track was investigated for trawl density and axis (direction) of trawling. For each minute (representing 80 m intervals) where the towfish was within 15 m of the seabed, latitude and longitude were noted, and the number of trawl marks across the swathe was enumerated and their primary axis noted (Figure 2). Where the run was more parallel than perpendicular to trawl marks, the count was made 80 m to either side of the axis of trawling at the 1 min position. Counts of trawl marks were transformed to density categories (Table 1), an approach that can take into account small inconsistencies in counts between different operators as well as differences in performance of the same operator over time. For two cases, this can be particularly important: first, depending on sediment type, some marks are very faint on sidescan and not easy to discern; second, in images with a very high incidence of marks, discerning individual marks can be difficult.
|
|
For each videotape or sledge track, every 5 min of tape, 2 min of video were examined using the anthropogenic and bioturbation scale shown in Table 1. This was 1 min either side of a 5 min-position record, representing sampling points with
150 m separation, i.e. 60 m viewed, followed by a 90 m interval. In addition, the density of the crinoid Leptometra phalangium was recorded on a similar categorized scale (0, absent, to 3, more than 20 per 2 min recording). The 5 min marks on video corresponded to the DGPS position marked during the recording.
A boundary box was defined for each of the categorized datasets around their spatial location. Within this boundary box, interpolation was carried out to create maps of the spatial distribution of the various categories of interest. Kriging was used to achieve the desired coverage while also allowing the interpolation method to be adjusted to the irregularly spaced data, and to quantify possible errors. Kriging works by creating predictions for locations, in this case grid cells, within the boundary box and based on the semi-variogram of the data and the spatial arrangement of measured values close to the location for which a prediction is being made. In general, interpolation using kriging is based on the statistical estimation of "local" values, weighted in relation to their distance from the sample point used, i.e. closer samples having more weight, and can be defined by the kriging estimator (T*):
|
| (2) |
Here, g1,..., gn are equivalent to the number of trawl marks detected, the trawl directions, etc., of the sampled points used for kriging. Clark (1979) writes that "This type of estimator is called a ‘linear’ estimator because it is a linear combination of the sample values", and "we can evaluate the accuracy of any linear estimator if we have a model for the semi-variogram", plus "we can produce a minimum variance unbiased linear estimator using the kriging technique, if we have a model of the semi-variogram". Standard-deviation-error maps were also created for each interpolation (not shown), and the final best kriging results were plotted as maps with various spatial-reference overlays.
A Spearman's rank correlation test allowed us to investigate the correlations between the measurements and ordinal data (categories) from the sidescan and video datasets. The correlations were undertaken first for the raw data along the video track (between features), then with abstracted interpolated data for the same geographic points for the sidescan-sonar data, and second for the spatially interpolated data within the area of overlap for the sidescan and video datasets.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Features noted from sidescan-sonar images
A variety of different bottom features could be seen on the sidescan images. Marks left by trawl doors were very evident in soft sediments, as furrows along the sediment surface. Sidescan images of trawling impacts in soft sediments viewed
200 m deep on the commercial trawling lane at Dia Island are shown in Figure 3. The images are screenshots from the sidescan tracks. Vertical, parallel scale bars in the images are at 20 m intervals, all images showing a swath of 196 m, 98 m to each side of the towfish. Gain settings between images are similar, all having been soft-filtered in processing (GeoPro LC software); no other image manipulation took place. Figure 3a shows an along-track image (track is along the trawling axis) of trawl-door marks with a towfish height 15 m above the seabed. Marks are visible at the edges of the image although they tend to fade at a distance of 80 m from the towfish. Trawl marks were relatively clear and countable. Figure 3b, taken in the same vicinity, shows similar marks, but the towfish is much closer to the sediment (5 m above). With the same gain settings, the trawl marks are not so evident >80 m from the towfish. However, they were clearer closer to the towfish.
|
Figures 3c and 3d show slant-range corrected images, the track under the towfish being closed-up, of trawling marks perpendicular to the axis of tow. At 10 m above the seabed, the cross-track marks are less evident than those along the track. Closer to the seabed (5 m), well-defined tracks, with deeper cuts and higher sediment spoil ramparts, are evident, but overall, the marks are less defined than in the along-track images.
Door marks were rarely noted to be paired, i.e. two continuous parallel marks, and in some cases herring-bone patterns were seen, where the door had obviously bounced and skidded over the seabed.
Sidescan images of trawling impacts in coarse sediments at 80 m deep at Gouves are shown in Figures 3e and 3f. Trawl marks were less evident in these images, and were seen as diffuse banding of the sediment surface. They were visible with the towfish quite high above the seabed (18 m), and there did not seem to be any difference between along-track and cross-track visibility. Because of the diffuse nature of the trawl marks, accurate enumeration from these images was not possible.
Wide-area impact mapping from sidescan sonar
Figure 4 shows the pattern of sidescan-sonar tracks across the commercial trawling lane at Dia Island. The lane is diagonally across the figure from northwest to southeast. An interpolated estimation of trawl-mark density after visual examination of the sidescan tracks can be seen in Figure 5, a density contour map after categorization on a scale of 0–5. Here, all counts below 2 are in the first, zero category, effectively filtering the data to indicate the principal axis of the trawling lane, and discarding data from the periphery.
|
|
The interpolated intensities and axes of trawl marks are shown in Figure 6. The main axis of trawling is northwest to southeast, following the depth contours and the edge of Dia Island, as well as the prevailing northwest wind direction. Coverage can be seen in the track plot in Figure 4. Trawling density was greatest towards the northwest and in the centre of the plot. Again, the principal axis of the lane is clear.
|
Features noted from video images
Around the commercial fishing lane at Dia Island, the towed video sledge generally gave clear images of the sediment, with the exception of occasional turbidity clouds that obstructed the view. This occurred when the system was towed near recent trawling activities. Sediments appeared generally soft with some slightly coarser, possibly harder, elements, with the appearance of maerl fragments, at the extreme northern edge of the area surveyed, corresponding to the bottom of the slope up to Dia Island.
Trawling impacts were readily evidenced through:
- Large furrow marks from trawl doors, seen as deep, continuous plough marks in the sediment surface, extending to at least 20 cm sediment depth, often triangular in section but asymmetrical, with a spoil heap mainly to one side of the mark (Figure 7a).
- Scrape marks from sweeps and trawl wires, evidenced as parallel sets of non-continuous scrape marks on the sediment surface, with small furrows and channels with a penetration into the sediment of a few centimetres (Figure 7a).
- Flattened areas from ground ropes and nets with no characteristic topographical surface features (Figure 7b).
- Areas with a soft appearance of the sediment surface, probably resettled suspended matter.
|
Fresh trawl marks had very clean-cut surfaces and irregularly shaped spoil heaps, with lighter coloured mud clasts. Older marks on the periphery of the ground, or those overlaid under fresh marks, were characterized by much smoother edges, with bioturbation holes in the side cuts of door marks, and new mounds on the features. In some areas flattened by the passage of trawl nets and ground ropes, old door marks appeared below the flattening level, with untouched biogenic features in them. Trawling marks were evident throughout the year, including in the closed season. Very fresh marks were less visible in the closed season, i.e. from the end of May to the end of September, indicating that they were biogenically weathered, but noticeable marks remained across the 4 month break.
Bioturbation features were noted at Dia Island primarily in the untrawled area, through the presence of mounds of various size, characteristic of thalasinnidean burrowing shrimps and echiuran worms (Figure 7c). Burrow entrances were also visible as vertical openings (characteristic of the burrowing shrimp Calocaris macandreae) and oblique openings (characteristic of the burrowing crab Goneplax rhomboides). On the sediment surface, the crinoid L. phalangium was evident generally in small aggregations (Figure 7d), the ophiuroid Ophiura ophiura either on the sediment surface or as buried star traces, along with the holothurian Stichopus regalis and occasionally the shrimp Parapenaeus longirostris. The burrowing goby, Leseurigobius friesii was also observed.
In the coarser-, harder-sediment, Gouves sampling area, trawl marks were not so visible on video. Larger door marks were evident as shallow scrapes 40 cm wide, with no strong traces of spoil heaps or sediment clasts, whereas small scrape marks were rarely visible. Maerl sand and maerl fragments tended to be aggregated within the scrape marks, making them more visible to the eye. Large patches of sediment appeared to be swept clean. The experimental trawl marks were only evident for 3–4 months.
In the Gouves untrawled area, more abundant features (not bioturbation features) were noted on the sediment surface, mostly in the form of sessile epifaunal species such as ascideans and sponges, bryozoans, hydroid masses, and polychaete tubes, but also as mobile forms: hermit crabs, asteroids, and ophiuroids. Maerl sand and larger maerl pieces were evident in algal complexes which were predominantly of the Lithamnion and Lithophyllum type, but also of soft, thick-leafed species.
Wide-area impact mapping from video
Figure 8 shows the tracks of individual video-sledge tows used in the analysis, and Figure 9 shows the interpolated spatial mapping of anthropogenic features on the scale of 0–3, scale defined in Table 1. The trawl lane is clear, depicted through the middle of the surveyed area by the darker shading, with low impact areas to its north and south. There is a general agreement with the trawling density mapped from sidescan, with some minor discrepancies.
|
|
Mapping the interpolated bioturbation impact factor in Figure 10 demonstrated very high levels of bioturation south of the trawl lane, modest levels to its north, and the lowest levels in the lane itself. This appears to be inversely related to that of anthropogenic impact, as shown in Figure 9. Bioturbation features were more pronounced to the south of the trawling lane than to the north, and matched the recording of coarser sediments at the base of the slope up to Dia island.
|
The distribution of the crinoid Leptometra phalangium (Figure 11) primarily south of the trawl lane matched that of the high bioturbation levels found there. Individual crinoids were only observed in the middle of the trawl lane (east side) during the closed trawling season, although they were found at the southern periphery of the trawl lane during the trawling season. The only individuals found north of the lane were on the far eastern edge, and none were observed at its northern periphery, at the bottom of the Dia Island slope.
|
The correlation coefficients for the tests between different features for raw and interpolated data are shown in Table 2. All correlations were highly significant. Positive correlations were noted between the trawling marks from video and sidescan, as well as between the density of crinoids and bioturbation. Negative correlations were noted between trawling marks and bioturbation and crinoid density. For the raw data, all correlation coefficients ranged from 0.2 to 0.5, whereas most of the correlations were stronger for the interpolated data. The highest coefficient values were found for sidescan against video trawling, bioturbation against video trawling, and bioturbation against crinoid density.
|
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Multiple-imaging techniques are now routinely used in habitat assessment to delimit areas of different bottom type (e.g. Germano et al., 1989; Smith and Greenhawk, 1998; Kostylev et al., 2001; Brown et al., 2002; Cutter et al., 2003). Of the few studies (Churchill, 1989; Riemann and Hoffmann, 1991; Schwinghamer et al., 1996, 1998; Service and Magorrian, 1997; DeAlteris et al., 1999; Friedlander et al., 1999; Humborstad et al., 2004; Malik and Mayer, 2007) using multiple methods (including imaging methods) to assess trawling impacts on the seabed, only Friedlander et al. (1999) and Malik and Mayer (2007) attempted to use quantitative imaging methodologies to assess the impact over wide areas. Although Malik and Mayer (2007) were not able to detect trawl marks with video in the same area that they had recorded marks with sonar systems, we found in common with the other studies that both sidescan sonar and a towed video sledge were useful in determining trawling impacts. We also demonstrated the added value of an ability to determine a categorized level of impact. The comparative characteristics and application to trawling impacts of the two methods used in this study are shown in Table 3. As a consequence of each technique’s tow speed, swathe, and resolution, they have differing degrees of capabilities, sidescan giving the wider area coverage and imaging trawl-door marks, whereas towed video, although covering a lesser swathe and area, was able to provide more detail on the passage of a trawl.
|
The trawling-impact intensity maps produced by the two methods generally agreed in terms of areas of overlap, with good correlation between estimates of trawling density from the two methods. Bioturbation levels and crinoid densities appeared to be negatively impacted by trawling, an observation supported by the correlation analysis. However, there were some minor inconsistencies between the maps. The seabed impacts of trawling are patchy, and bioturbation is not expected to be completely absent from impacted areas. Some of the differences between the features may be due to us not taking into account the layback of the video gear when position fixing, so reducing accuracy in geo-referencing the video data. It is also possible that sidescan sonar detects sub-bottom disturbances, and particularly older disturbances, in the sediment that may not be visible from videos of the sediment surface. Malik and Mayer (2007) attributed the non-detection of trawl marks by video (as detected from sonar) to uncertainties in the positioning and the sensitivity or orientation, or both factors, of their camera. They surmised that the small field of view pointing vertically downwards led to difficulties in resolving topographical features.
With adequate survey time and good geo-referencing, it is straightforward to build up 100% coverage of an area with parallel sidescan tracks, something that is almost impossible with a towed video because of its narrow field of view. Therefore, any full area analysis with video could only be undertaken using interpolation between the tracks in a manner similar to that in which acoustic ground-discrimination sonar data are used. Because of operational constraints during the data collection of both the sidescan and underwater-video data, the spatial distribution of the resultant categorizations is irregular, some areas of trawling lanes being more densely sampled than others. To improve the accuracy of spatial-interpolation techniques, the underlying data should be as densely and equally sampled over the target area as possible. Therefore, results from parts of the study areas that were sparsely sampled should be treated with caution.
In the study by Friedlander et al. (1999), direct trawl counts were used to estimate density. They noted that their estimates were minimum ones, owing to biases including variability in resolution, obliteration of older marks by more recent trawling, burial by sediment, or natural physical reworking. Malik and Mayer (2007), using high-level filtering and enhancement algorithms, detected trawl marks from multibeam records, and overlaid them on area maps. Fished and non-fished areas can be observed on wide-area maps and the density compared visually. Here, category counts were used to improve uncertainty, particularly from differences between observers and in high-density areas where underlying marks might have been obscured. Resolution was poor in some areas, particularly where the towfish was towed too high above the seabed; in such cases, some data were excluded from the analyses, although the area is covered by data interpolation.
From the point of view of shadow and the consequent ability to visualize the marks, the direction of approach is important to area surveying. Friedlander et al. (1999) found a significant difference in the detection of marks, fewer being detected when approached at 90° to the transect path. They recommended that transects be subparallel to the direction of trawling. We believe that a 90° track should be avoided and that, if there is less than 100% coverage of an area, tangential transects could be undertaken. Video transects are better in a 90° or tangential direction because, if run parallel, one trawl mark might be in view continuously, giving an estimate of 100% coverage, although it might be the only mark present.
The selection of the timed observation period and the interval between periods (2 min viewed every 5 min, representing 60 m viewed, followed by a 90 m interval) for analysing video material worked well in this study. It is, however, ground- or feature-dependent. The period to be viewed depends on the requirements of the investigation, i.e. on what features are being looked for in a discrete sampling area. The advantage of a longer period is the greater resulting representation of seabed conditions in an area. However, its disadvantage is in the variety of different features that may be noted, from a clear area to a freshly trawled area in the middle of a trawl ground, leaving the observer in a dilemma as to which category to use in defining an observation. On the other hand, shorter periods may miss important features unless the interval between them is short: ideally if review time is not a limiting factor, then the observation periods should be continuous.
Humborstad et al. (2004) detected parallel door marks from the same trawl, 140 m apart. At the start of the present study, we thought that it would be easy to identify paired doors. However, such observations were rare in our data, probably because door spread for the Greek commercial trawl is quite narrow (20–30 m), and door marks are not easy to differentiate at high density. However, there are a number of factors which, in combination, may lead to the lack of detection of paired marks, including:
- one of the door pairs may have been out of the sidescan view;
- the two individual marks from a pair may be imaged differently when viewed from one side. A mark of a trawl door is asymmetrical (Smith et al., 2000; Humborstad et al., 2004): a door can dig deeper on one side of the plough mark and push excavated material in towards the trawls, so the door mark may have different shadow patterns when viewed from one side or the other;
- one of each of the pair of doors may have been on different sides of the sidescan, and with slightly different gain settings to the transducers, one mark may not have been so apparent.
- the rigging of the trawling gear may have been uneven such that one door was not in constant contact with the sediment or was digging in deeper.
- the trawler may have been turning and one door may not have been in contact with the seabed.
It is obvious that trawl doors will dig deeper into softer sediments, and less into coarser and more compact sediments. Certainly in the Mediterranean, shallower areas (<100 m deep) are predominantly characterized by coarser, loose sediments. With less binding agents in the sediment, any furrows or sediment berms will not hold their shape and will be more easily weathered by water movement. Trawl marks observed at the coarse-sediment Gouves site in sidescan-sonar images were very diffuse. It is therefore likely that these shallow marks, visible on the sonar images, had been enhanced by the accumulation of coarser fragments (maerl) in the shallow door scrapes. These fragments would have a higher reflectivity, showing up as dark streaks in both along- and cross-track images. Further evidence of this was the fact that the trawl marks were equally visible when the towfish was well above the seabed.
With the greater resolution of video, relative ageing of trawl marks may be seen and estimated quantitatively through categorization. This is not possible with sidescan sonar, where for a strongly marked scar, one cannot distinguish between an old persistent mark that had dug deeply into the sediment and a fresher mark that had not. One problem with video, however, is that it is susceptible to poor visibility, sediment clouds obscuring the image and data being lost along a continuous track. Sonar images are not affected by turbidity.
The greater resolution of the video also allowed other features of importance to be seen, including lesser anthropogenically introduced marks, e.g. flattening and scrape marks, as well as biological features, such as the presence and abundance of identifiable fauna. Leptometra phalangium in the Aegean is a relatively large species, and was present often enough to estimate densities. Large, visible, and identifiable fauna are, however, sparsely distributed in the oligotrophic southern Aegean, with L. phalangium being the area exception. Its distribution was well correlated with higher levels of bioturbation and it had a significant negative correlation with trawling impact. It is a low motile, fragile species, and it has been identified as a key indicator species for unimpacted areas (Smith et al., 2000). Density indicators could easily be applied to other relatively large sessile epifauna in other areas, particularly sponges, sea pens, soft corals, and ascideans. Although the bioturbating community was not identified to species level, the indicator is useful with respect to trawling impact, as noted by the largely inverse representation between the trawl-mark density and bioturbation maps. Trawling will impact bioturbation activity negatively through removal, indirect mortality (damage, exposure of deep burial), and disturbance of bioturbators, particularly those close to the sediment surface.
The two techniques are rapid assessment tools for evaluating trawling impacts. They avoid the complexity of traditional benthic monitoring which, although it provides a wealth of data, is extremely processing-intensive and consequently expensive, and is often complex and difficult to interpret. However, the two methods are complementary, and together provide added value to traditional sampling programmes. The methods developed here are highly suitable for quick assessment for management purposes, and may be used to ground-truth vessel-monitoring-system (VMS; the EU’s current satellite-tracking system for fishing vessels) data or to link directly to such data to investigate the microscale distribution of trawling, e.g. how trawling is concentrated at a particular time or changes between time periods. Having an understanding of microscale distribution is very important in evaluating area impacts more precisely (Rijnsdorp et al., 1998; Piet et al., 2000; Murawski et al., 2005). Although accurate and of high resolution, VMS data take some time to collate, and the systems currently in use do not cover all areas or all fleet and vessel segments. Imaging methods give the opportunity to observe similar data in one snapshot.
There is certainly room for improvement in the use of imaging technologies, especially concerning speed, precision, and accuracy in data acquisition and analysis. Indeed, we are extending data collection in systematic sonar surveys (100% coverage) and are involved in the development of automated computer detection of trawl marks from sidescan-sonar records.
|
Acknowledgements |
|---|
We thank the captain and crew of RV "Philia" for their assistance in the collection of data at sea. The methodologies were developed under discussion with project partners Roger Coggan of Cefas, UK, and R. J. A. Atkinson of UMBSM, UK. We developed the ideas and techniques while financed by the following EU projects: comparison of rapid methodologies for quantifying environmental impacts of otter trawls (DGXIV Study Project 98/017), COST-IMPACT: costing the impact of demersal fishing on marine ecosystem processes and biodiversity (FP5 QoL Q5RS-2001-00 993) and AMASON: advanced mapping with sonar and video (FP5 EESD: EVK3/CT/2001/00 059).
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Brown C. J., Cooper K. M., Meadows W. J., Limpenny D. S., Rees H. L. Small scale mapping of sea-bed assemblages in the eastern English Channel using sidescan sonar and remote sampling techniques. Estuarine, Coastal and Shelf Sciences (2002) 54:263–278.[CrossRef]
Churchill J. H. The effect of commercial trawling on sediment resuspension and transport over the Middle Atlantic Bight continental shelf. Continental Shelf Research (1989) 9:841–864.[CrossRef][Web of Science]
Clark I. Practical Geostatistics. (1979) London: Applied Science Publishers. 119. pp.
Collie J. S., Hall S. J., Kaiser M. J., Poiner I. R. A quantitative analysis of fishing impacts on shelf sea benthos. Journal of Animal Ecology (2000) 69:785–798.[CrossRef][Web of Science]
Cutter G. R., Rzhanov Y., Mayer L. A. Automated segmentation of seafloor bathymetry from multibeam echosounder data using local Fourier histogram texture features. Journal of Experimental Marine Biology and Ecology (2003) 285/286:355–370.
Dayton P. K., Thrush S. F., Agardy M. T., Hofman R. Environmental effects of marine fishing. Aquatic Conservation (1995) 5:205–232.[CrossRef]
DeAlteris J., Skobe L., Lipsky C. The significance of seabed disturbance by mobile fishing gear relative to natural processes: a case study in Narragansett Bay, Rhode Island. American Fisheries Society Symposium (1999) 22:224–237.
Friedlander A. M., Boehlert G. W., Field M. E., Mason J. E., Gardener J. V., Dartnell P. Sidescan-sonar mapping of benthic trawl marks on the shelf and slope off Eureka, California. Fishery Bulletin US (1999) 97:786–801.
Germano J. D., Rhoads D. C., Boyer L. F., Menzie C. A., Ryther J. A. REMOTSR imaging and sidescan sonar: efficient tools for mapping sea floor topography, sediment type, bedforms, and biology. In: Oceanic Processes in Marine Pollution. Scientific Monitoring Strategies for Ocean Waste Disposal, 4—Hood D. W., Schoener A., Park P. K., eds. (1989) Malabar, Florida: R.E. Krieger Publishing. 39–48.
Gislason H., Sinclair M., Sainsbury K., O’Boyle R. Symposium overview: incorporating ecosystem objectives within fisheries management. ICES Journal of Marine Science (2000) 57:468–475.
Hall S. J. The Effects of Fishing on Marine Ecosystems and Communities. (1999) Oxford: Blackwell Science. 274. pp.
Hiddink J. K., Jennings S., Kaiser M. J. Assessing and predicting the relative ecological impacts of disturbance on habitats with different sensitivities. Journal of Applied Ecology (2007) 44:405–413.[CrossRef][Web of Science]
Humborstad O-B., Nottestad L., Lokkeborg S., Rapp H. T. RoxAnn bottom classification system, sidescan sonar and video-sledge: spatial resolution and their use in assessing trawling impacts. ICES Journal of Marine Science (2004) 61:53–63.
Jennings S., Kaiser M. J. The effects of fishing on marine ecosystems. Advances in Marine Biology (1998) 34:201–352.[Web of Science]
Kaiser M. J., Clarke K. R., Hinz H., Austen M. C. V., Somerfield P. J., Karakassis I. Global analysis and prediction of the response of benthic biota to fishing. Marine Ecology Progress Series (2006) 311:1–14.[CrossRef][Web of Science]
Kaiser M. J., Collie J. S., Hall S. J., Jennings S., Poiner I. R. Modification of marine habitats by trawling activities: prognosis and solutions. Fish and Fisheries (2002) 3:114–136.[CrossRef]
Kenny A. J., Cato I., Desprez M., Fader G., Schuttenhelm R. T. E., Side J. An overview of seabed-mapping technologies in the context of marine-habitat classification. ICES Journal of Marine Science (2003) 60:411–418.
Kostylev V. E., Todd B. J., Fader G. B. J., Courtney R. C., Cameron G. D. M., Pickrill R. A. Benthic habitat mapping on the Scotian Shelf based on multibeam bathymetry, surficial geology and sea-floor photographs. Marine Ecology Progress Series (2001) 219:121–137.[CrossRef][Web of Science]
Malik M. A., Mayer L. A. Investigation of seabed fishing impacts on benthic structure using multi-beam sonar, sidescan sonar, and video. ICES Journal of Marine Science (2007) 64:1053–1065.
Murawski S. A., Wigley S. E., Fogarty M. J., Rago P. J., Mountain D. G. Effort distribution and catch patterns adjacent to temperate MPAs. ICES Journal of Marine Science (2005) 62:1150–1167.
Piet G. J., Rijnsdorp A. D., Bergman M. J. N., van Santbrink J. W., Craeymeersch J., Buijs J. A quantitative evaluation of the impact of beam trawling on benthic fauna in the southern North Sea. ICES Journal of Marine Science (2000) 57:1332–1339.
Pinn E. H., Robertson M. R., Shand C. W., Armstrong F. Broad-scale benthic community analysis in the Greater Minch area (Scottish West Coast) using remote and nondestructive techniques. International Journal of Remote Sensing (1998) 19:3039–3054.[CrossRef][Web of Science]
Riemann B., Hoffmann E. Ecological consequences of dredging and bottom trawling in the Limfjord, Denmark. Marine Ecology Progress Series (1991) 69:171–178.[CrossRef][Web of Science]
Rijnsdorp A. D., Buys A. M., Storbeck F., Visser E. G. Micro-scale distribution of beam trawl effort in the southern North Sea between 1993 and 1996 in relation to the trawling frequency of the sea bed and the impact on benthic organisms. ICES Journal of Marine Science (1998) 55:403–419.
Schwinghamer P., Gordon D. C., Rowell T. W., Prena J., McKeown D. L., Sonnichsen G., Guigné J. Y. Effects of experimental otter trawling on surficial sediment properties of a sandy-bottom ecosystem on the Grand Banks of Newfoundland. Conservation Biology (1998) 12:1215–1222.[CrossRef][Web of Science]
Schwinghamer P., Guigne J., Siu W. Quantifying the impact of trawling on benthic habitat structure using high resolution acoustics and chaos theory. Canadian Journal of Fisheries and Aquatic Sciences (1996) 53:288–296.
Service M., Magorrian B. H. The extent and temporal variation of disturbance to epibenthic communities in Strangford Lough, Northern Ireland. Journal of the Marine Biological Association of the UK (1997) 77:1151–1164.
Shand C. W., Priestley R. A towed sledge for benthic surveys. Scottish Fisheries Information Pamphlet (1999) 22/1999.:8. pp.
Smith C. J., Papadopoulou K-N., Diliberto S. Impact of otter trawling on an eastern Mediterranean commercial fishing ground. ICES Journal of Marine Science (2000) 57:1340–1351.
Smith C. J., Rumohr H. Imaging techniques. In: Methods for the Study of Marine Benthos—Eleftheriou A., McIntyre A., eds. (2005) Oxford: Blackwell Science. 87–111. 418 pp.
Smith C. J., Rumohr H., Karakassis I., Papadopoulou N. Analysing the impact of bottom trawls on sedimentary seabeds with sediment profile imagery. Journal of Experimental Marine Biology and Ecology (2003) 285/286:479–496.
Smith G. F., Greenhawk K. N. Shellfish benthic habitat assessment in the Chesapeake Bay: progress toward integrated technologies for mapping and analysis. Journal of Shellfish Research (1998) 17:1433–1437.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  V. Stelzenmuller, S. I. Rogers, and C. M. Mills Spatio-temporal patterns of fishing pressure on UK marine landscapes, and their implications for spatial planning and management ICES J. Mar. Sci., September 1, 2008; 65(6): 1081 - 1091. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||