Journal of Experimental Marine Biology and Ecology 255 (2000) 111±129 www.elsevier.nl/locate/jembe

An in situ study of predator aggregations on ( maximus (L.)) dredge discards using a static time-lapse camera system

L.O. Vealea,* , A.S. Hill b , A.R. Brand a aUniversity of Liverpool Port Erin Marine Laboratory, Port Erin, Isle of Man IM96JA, UK bSEPA, Clearwater House, Heriot-Watt Research Park, Edinburgh EH14 4AP, UK Received 25 May 2000; received in revised form 19 September 2000; accepted 21 September 2000

Abstract

The impact of demersal ®shing gears on benthic habitats and has been the subject of much attention recently, and suggestions have been made that scavenging epifaunal species may bene®t at the population level from the additional food source provided by discards. This paper investigates some aspects of this process, including the relative attractiveness to predators of different discard species, and the role of damage in scavenger attraction. A time-lapse video system with a 1000 m long cable was positioned in an area closed to ®shing, adjacent to the most heavily ®shed scallop () ground in the . A variety of undamaged and damaged by-catch were positioned in front of the camera, and the subsequent predator aggregations investigated. Densities of scavenger species up to 200 times that of the background population were observed, and aggregations of some species persisted for up to 3 days. The most frequently recorded scavengers, and therefore presumably those species most likely to bene®t from discards as a food source, were: Asterias rubens L., irregularis (Pennant), Liocar- cinus spp Stimpson, Pagurus spp Fabricius and Callionymus lyra L. Predator attraction to apparently undamaged queen , opercularis (L.), was almost as high as to damaged A. opercularis. Of all the prey species studied, queen scallops were the most attractive to scavengers. A directional relationship was found between the ambient water current and the arrival of the star®sh, Asterias rubens.  2000 Elsevier Science B.V. All rights reserved.

Keywords: Fisheries by-catch and discards; Fishing impact; Pecten maximus; Predator aggregation; Time-lapse video

*Corresponding author. Tel.: 144-1624-831-017; fax: 144-1624-831-001. E-mail address: [email protected] (L.O. Veale).

0022-0981/00/$ ± see front matter  2000 Elsevier Science B.V. All rights reserved. PII: S0022-0981(00)00295-1 112 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

1. Introduction

Mobile demersal ®shing gears are known to have a detrimental effect on benthic epifaunal and infaunal communities (Dayton et al., 1995; Thrush et al., 1995; Jennings and Kaiser, 1998), often dramatically increasing local mortality in the wake of the gear (Kaiser and Spencer, 1994b); this may have long-term implications for the structure of the community (Thrush et al., 1998). Both those animals disturbed or damaged by the passage of the gear and left on the seabed, and those retained in the catch and subsequently discarded, are known to attract mobile predators (Kaiser and Spencer, 1994a, 1996), which act as facultative scavengers (Britton and Morton, 1994). This ability to feed opportunistically on dead and dying animals, coupled with robustness to capture and damage in the ®shing gear, may confer an increased survivorship on certain species that, in turn, may lead to an enhanced population size (Polis et al., 1996; Ramsay et al., 1997b), e.g. the ¯at®sh, Limanda limanda (L.) (Kaiser and Ramsay, 1997). In intensively ®shed areas, such as the , this carrion will inevitably subsidize some marine food webs (Furness, 1996; Ramsay et al., 1997b). The levels of ®shing effort presently exerted by scallop (Pecten maximus) and (Aequipecten opercularis) dredging in the northern Irish Sea (approx. 177,000 metre hours for the 1994±95 ®shing season) will potentially introduce a large amount of by-catch material [approximately 231 tonnes for the 1994±95 season (L. Veale, unpublished data)] into the food web; a proportion of this will become available to benthic predators and scavengers. Aggregations of scavenging species after the passage of towed demersal ®shing gears have been recorded in several previous studies, mainly by submersible, video or diver observations made along a ®shed track (e.g., Medcof and Bourne, 1964; Caddy, 1973; Chapman et al., 1977; Murawski and Serchuk, 1989; Kaiser and Spencer, 1994b). These have all noted increased densities of several ®sh and invertebrate scavenging species in response to ®shing activities, but details such as the relative attractiveness of the different prey species, the importance of prey damage, and the relationship between direction of attraction and water movement, have not been addressed. It is of particular importance to ascertain whether apparently undamaged animals are equally likely to be preyed upon after discard, as this will have implications for the potential bene®ts of gear modi®cations designed to reduce damage. Some studies have used baited time-lapse stills cameras deployed from research vessels to investigate the scavengers attracted to damaged animals typical of material discarded from ®shing gears (e.g., Kaiser and Spencer, 1996; Ramsay et al., 1997b). Here, the maximum scavenger activity occurred within 24 h. This approach allows a detailed investigation of the arrival times of different scavenging species and the rate of bait consumption. However, a stills camera will inevitably return fewer frames than a video camera, and its deployment from a ship will either limit the duration of the investigation, or incur high ship-time costs. Invertebrate scavenger species are generally slow moving, and slow at ingesting and digesting food, so it is important that such investigations are of suf®cient duration to adequately record the arrival of all species attracted to the prey, and to monitor their subsequent departure. A static time-lapse video system was deployed in this study to identify the major L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 113 species involved in scavenging dead and damaged benthos typical of that either discarded from scallop dredges, or damaged and left on the seabed. A 1000 m long cable allowed the camera to be positioned offshore, on an area of seabed typical of that supporting the local scallop ®shery, but within an area closed to commercial ®shing (Bradshaw et al., 2000). Recordings totalling over 2000 h were made, which would have been impossible to achieve using SCUBA divers or cameras deployed from a ship. These extended time periods ensured that the aggregation and dispersion of slow-moving invertebrate scavengers were adequately monitored. Firstly, the relative merits of using white and red light to record nighttime footage were compared. Next, the aggregation of scavengers on mixed damaged benthos was examined, and then the attractiveness of different by-catch species was investigated using four mono-speci®c baits: damaged Aequipecten opercularis, Pecten maximus, Asterias rubens and undatum L. These species were chosen as they represented some of the most abundant members of the catch assemblage of the north Irish Sea scallop ®shery (Veale et al., 2000). The composition and size of scavenger aggregations attracted to both damaged and undamaged A. opercularis were then compared. An exploration of the interaction of water current direction and distribution of olfactory stimuli was conducted for Asterias rubens.

2. Materials and methods

From the 14th June to 5th October 1996, a static video camera system was deployed in the area closed to ®shing off Port Erin, Isle of Man. This was during the closed season for the great scallop, which runs from 1st June to 31st October inclusive: no commercial dredging occurred in the vicinity during the study. A Rovtech Systems low-light colour camera with two 250 W lights was used. The signal was transmitted to a terrestrial control box via a booster unit and 1000 m of cable. The image was recorded on a Panasonic SVHS time-lapse video recorder, and real time video was viewed on a JVC colour monitor. The camera was situated approximately 600 m offshore in 25 m depth, due west of the Marine Laboratory (Fig. 1), mounted on a galvanised Dexian frame at an angle of 458, 1 m above the seabed. The camera was orientated facing east, perpen- dicular to the prevailing currents in the area, which travel north or south, depending on tidal state. Throughout the study period divers positioned a variety of baits loose on the seabed in front of the camera. Mixed bait comprised a variety of dredge-caught epifaunal species, including Aequipecten opercularis, Pecten maximus, antiqua (L.), , Echinus esculentus Lamarck, and Asterias rubens, which were damaged just prior to baiting: damage was applied with stones and metal bars to simulate that observed in damaged by-catch animals. Baited periods were interspersed with unbaited (control) periods, and the camera was repositioned three times to reduce the number of animals permanently associating themselves with the frame, although no differences in background (unbaited) abundance were noted after repositioning. Initially, the relative merits of using white and red light to record nighttime footage were compared in single 3-day unbaited trials with each type of light. Then the 114 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 relocated over three positions throughout the study. Fig. 1. Approximate positions of the ®xed camera within the area closed to ®shing. Diamonds indicate the positions of navigation buoys. The camera was L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 115 aggregations on mixed damaged benthos were investigated, with two unbaited/baited comparisons. Mono-speci®c baits of Aequipecten opercularis, Pecten maximus, Asterias rubens and Buccinum undatum were also studied in single unbaited/baited trials, and then ®nally the effect of damage incurred by Aequipecten opercularis on the ensuing aggregation was investigated by a single undamaged/damaged comparison. Generally, 3-day unbaited periods were followed by 3-day baited periods; previous studies in the Irish Sea had noted that peak numbers occurred less than 24 h after baiting (Kaiser and Spencer, 1996). The extended time period used here allowed the dispersion of the aggregation to be monitored. On one of the mixed baiting occasions, the recording was continued for 5 days after baiting, to further assess the dispersal of the aggregation. The direction of arrival of Asterias rubens at all the baits described above (where possible) was compared with simultaneous estimates of water current direction, using circular statistics (Batschelet, 1981). The current direction was noted by observing the movement of particles in the water, or pieces of seaweed across the seabed, and the direction of arrival of A. rubens was estimated as it entered the ®eld of view of the camera. The background densities of the major scavenging species were estimated by sixteen 200 m2 diver transect surveys. Pairs of divers followed a marked 50 m bottom line (repositioned for each survey), recording animals within 2 m of the line on either side. The width of the survey was delimited by a 4 m pole, on which were mounted slates for recording data.

2.1. Data analysis

The video tapes were played back on the same Panasonic SVHS time-lapse recorder, which was equipped with an on-screen date and time display. Instantaneous counts of all the species present were made every 10 min of real time. It was impossible to determine whether or not some species (e.g., star®sh) were feeding on the bait, so counts include all animals in the vicinity of the bait, but not necessarily feeding on it. For graphical presentation, hourly mean abundance of each species at the bait was calculated and plotted against time. Mean abundance was used in favour of the maximum because it included elements of animal abundance and residence time at the bait. Due to lack of independence between consecutive observations in a time series, it was decided to use daily (24 h) means for statistical analysis. This time period was large enough to ameliorate the independence problem, and also eliminated any diel differ- ences. However, it meant that there were only three replicates for each treatment. A square-root transformation provided an approximation to normality so that parametric ANOVA could be used to compare daily mean numbers of each species: before and after baiting, under the different light regimes, or between damaged and undamaged A. opercularis. After the second mixed baiting recording was continued for 5 days. On this occasion, three groups were used in the ANOVA: days 1±3 before baiting; days 1±3 after baiting; days 4±5 after baiting. Multivariate analysis was used to assess any differences between the assemblages observed daily in front of the camera, before and after baiting with mixed damaged 116 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 benthos. The data were standardised and square-root transformed; no species were excluded. These decisions were based on the fact that only a small number of species were visible in front of the camera, and no species greatly dominated the assemblage. Dendrograms and non-metric multidimensional scaling (MDS) plots of the sample relationships de®ned by the Bray±Curtis similarity index were calculated and plotted using the PRIMER software package. The statistical signi®cance of any difference observed before and after baiting was tested using ANOSIM (Clarke and Warwick, 1994). The relationship between the direction of arrival of Asterias rubens at the bait, and the direction of tidal water ¯ow, was examined using the Jupp±Mardia circular±circular correlation coef®cient (Batschelet, 1981). The test statistic nr 2 is compared with tabulated values for x 2 with four degrees of freedom. However, there is no way of distinguishing a positive from a negative correlation using this technique, so the differences between the directions from which the current and the A. rubens were coming were calculated. It was postulated that if A. rubens were moving into the water current, possibly in response to olfactory stimuli, then this mean angle (6con®dence interval) would fall between 135 and 2258 (i.e. 1806458). The con®dence interval for the population mean angle (u ) was derived from graphed values presented in Batschelet (1981).

3. Results

A total of 22 taxa were identi®ed throughout the study period (Table 1). They ranged in percentage occurrence (number of frames in which the animal was observed/total number of frames examined*100) from 0.01 to 45.2%. The star®sh, Asterias rubens, was the most frequently observed species, appearing in three times as many frames as any other species. The next most frequently observed species were: the , Liocarcinus spp and Pagurus spp; the star®sh, ; and the demersal ®sh, Callionymus lyra. All other species were observed at much lower frequencies. Initially, the relative effects of red and white light were assessed; 3-day periods of each light regime, without bait, were ®lmed. Statistical analysis of the daily means showed signi®cantly elevated counts for Pagurus spp, Asterias rubens, Liocarcinus spp and Astropecten irregularis under white light conditions compared to red light. A. irregularis was also signi®cantly more abundant at night than during the daylight hours (Table 2, Fig. 2). Additionally, L. and A. irregularis both exhibited a signi®cant interaction term between type of light and time of day (Table 2), i.e. activity over the day/night cycle was related to the type of light used; at night they were more frequently observed under white light. Given the signi®cance of these results, red light was used for all subsequent studies, and the data collected under white light was not used. Increases in abundance of Asterias rubens, Astropecten irregularis, Pagurus spp, Liocarcinus spp and the other species pooled, were apparent after baiting with mixed damaged benthos (Figs. 3 and 4). Aggregations of up to 17 individual animals were visible at any one time. The maximum abundance of several species aggregating on the L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 117

Table 1 Taxa observed throughout the duration of the study, and their percentage and rank occurrence in all video frames examined Phylum Species/taxa % Rank occurrence occurrence Crustacea Atelecyclus rotundatus 0.2 15.5 Cancer pagurus 4.1 7 Inachus spp 0.3 12 Liocarcinus spp 15.5 3 Pagurus spp 17.1 2 Eledone cirrhosa 0.1 18 Nudibranch 1 0.1 18 Nudibranch 2 0.3 12 Nudibranch 3 0.3 12 Echinodermata Asterias rubens 45.2 1 Astropecten irregularis 12.3 4 Crossaster papposus 0.02 21 Porania pulvillus 1.9 8 Ophiocomina nigra 0.2 15.5 Ophiura spp 0.3 12 Ophiothrix fragilis 0.3 12 Pisces Agonus cataphractus 0.5 9 Callionymus lyra 10.9 5 Flat®sh 4.5 6 Synagnathus sp 0.1 18 Scyliorhinus caniculata 0.01 22 Triglidae 0.05 20 bait within the 2 m2 ®eld of view was many times greater than the background densities recorded by diver surveys in the same area at the same time (Table 3). Some species dispersed more quickly than others. For example, A. rubens declined slowly from the initial mean of six individuals present, in the days following baiting, reaching the background level by day 4, but aggregations of Liocarcinus spp were much shorter lived, lasting only 12 h (Fig. 3). This is inevitably linked to the time required for

Table 2 Comparison between the daily mean numbers of the six most common species observed under full and red light conditions (no bait used) (two-way ANOVA). *Signi®cant at 5% level, **signi®cant at 1% level. Results in bold type had a power of the performed test of .0.8 Species Light used Time of day Interaction (white or red) (day or night) (Light3Time of day) Asterias rubens F 5 9.5, P , 0.05* F 5 0.6, P 5 0.47 F 5 0.1, P 5 0.71 Astropecten irregularis F 5 89.0, P , 0.01** F 5 49.7, P , 0.01** F 5 40.0, P , 0.01** Cancer pagurus F 5 1.1, P 5 0.33 F 5 1.1, P 5 0.33 F 5 17.9, P , 0.01** Liocarcinus spp F 5 6.3, P , 0.05* F 5 4.8, P 5 0.07 F 5 0.6, P 5 0.46 Callionymus lyra F 5 0.1, P 5 0.78 F 5 1.3, P 5 0.29 F 5 0.3, P 5 0.60 Pagurus spp F 5 22.0, P , 0.01** F 5 0.0, P 5 0.95 F 5 0.0, P 5 0.95 118 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

Fig. 2. Hourly mean numbers of animals observed over unbaited 3-day periods with full light (top) and red light (bottom). Night and day are indicated by the black and white bar. *No data recorded during this night. feeding, and their relative mobilities. It is also clear that A. irregularis is a night-active species. Statistical analysis of the data from the ®rst baiting occasion revealed signi®cant differences between daily mean abundance before and after baiting, for Asterias rubens

(ANOVA, F1,4 5 18.61, P , 0.05), Astropecten irregularis (ANOVA, F1,4 5 8.24, P , 0.05), Cancer pagurus (ANOVA, F1,4 5 22.47, P , 0.01) and the data for Pagurus spp L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 119

Fig. 3. Mean numbers of animals observed per hour before and after the ®rst baiting event. The solid vertical line indicates the baiting event, and the dotted lines enclose areas of the graphs where no data were recorded. Night and day are indicated by the black and white bars. 120 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

Fig. 4. Mean numbers of animals observed per hour before and after the second baiting event. The solid vertical line indicates the baiting event, and the dotted lines enclose areas of the graphs where no data were recorded. Night and day are indicated by the black and white bars. L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 121

Table 3 Maximum densities of some species observed at the bait within the 2 m2 ®eld of view, compared to background density estimates obtained from 16 diver surveys in the area at the time Species Maximum density Mean back- Size of observed at bait ground density aggregation (Nos. per 2 m22 (Nos. per 2 m ) (3 increase) ®eld of view) Asterias rubens 8 0.18 44 Astropecten irregularis 8 0.04 200 Pagurus spp 7 0.09 78 Liocarcinus spp 4 0.10 40 Callionymus lyra 5 0.07 70

were signi®cant at the 10% level (ANOVA, F1,4 5 6.16, P 5 0.068). The second baiting occasion showed signi®cant differences only for A. rubens (ANOVA, F2,5 5 16.71, P , 0.01), and ¯at®sh (ANOVA, F2,5 5 5.91, P 5 0.05). Subsequent Tukeys multiple range comparison identi®ed differences between the group comprising days 1±3 after baiting and both the other two groups (days 1±3 before and days 4±5 after baiting) for A. rubens, demonstrating that the aggregation has dispersed after 3 days. Multivariate analysis showed a clear separation between the assemblages present before and after baiting (e.g., the second baiting with mixed damaged benthos Ð Fig. 5). Notably, days 4 and 5 after baiting grouped together with the days before baiting, indicating a gradual shift back towards the pre-baiting structure, with the ®rst day after baiting (a1) the most distant from the pre-baiting samples (b1±3) (Fig. 5). A signi®cant difference in assemblage structure before and after baiting was demonstrated for this baiting (ANOSIM: Global R 5 0.57, P 5 0.036) where groups comprised days 1±3 before, days 1±3 after, and days 4±5 after baiting. No signi®cant changes in scavenger assemblages were found for the other seven trials. To assess the effects of damage to discards on the aggregation of scavengers, undamaged and chipped (minor sub-lethal damage) queen scallops (n 5 15), Aequipecten opercularis, were used as bait in separate 3-day periods. Asterias rubens again dominated the scavenging assemblage (|50% of all animals observed). The immediate increases in A. rubens abundance (day 1) are more or less equal for both damaged and undamaged baits, but they appeared to remain at the damaged bait longer over the whole 3-day period (Fig. 6). There was no signi®cant difference between A. rubens abundance on damaged and undamaged queens, but there was a signi®cant increase over unbaited abundance after baiting with damaged queens (ANOVA: F1,4 5 9.79, P , 0.05). Both Pagurus spp and Cancer pagurus abundances increased signi®cantly after baiting with both damaged and undamaged queens (ANOVA: F2,65 157.4, P , 0.01 and F 2,6 5 810.3, P , 0.01, respectively), but there was no difference between aggregations on the two different damage grades. Only Callionymus lyra demonstrated a signi®cant difference between abundance on damaged and undamaged queens, in addition to increases over unbaited periods (ANOVA: F2,6 5 18.1, P , 0.05), being more abundant after baiting with damaged queens. This species also showed a marked periodicity in activity, being most active in the middle of the day. 122 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

Fig. 5. Dendrogram and non-metric MDS plot showing the relationships between assemblages observed before (b) and after (a) the second baiting with mixed damaged benthos. Data have been standardised, square-root transformed, and no species have been excluded. The dotted lines indicate samples which are similar at the 75% level. L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 123 me of baiting. Night ) (left) and those with Aequipecten opercularis observed after baiting with undamaged queens ( Callionymus lyra and Asterias rubens Fig. 6. Hourly mean numbers of minor damage (shell chipping) (right).and The day horizontal lines are indicate indicated the by unbaited average the from black the and ®rst baiting white study. bar. The vertical lines show the ti 124 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

Table 4 Summary of results of experiments using baits comprising different by-catch species to attract scavengers to a baited video camera Bait Scavenging species showing signi®cant increases Mixed benthos Asterias rubens, Astropecten irregularis, Pagurus spp, Cancer pagurus, ¯at®sh Aequipecten opercularis Asterias rubens, Pagurus spp, Cancer (damaged and undamaged) pagurus, Callionymus lyra Pecten maximus Asterias rubens, Astropecten irregularis, (undamaged) Pagurus spp Asterias rubens Flat®sh Buccinum undatum Callionymus lyra, Liocarcinus spp, ¯at®sh

Aggregation on undamaged animals was further demonstrated when the great scallop, Pecten maximus, was used as bait. After baiting, signi®cant increases in abundance were found for Asterias rubens (ANOVA: F1,4 5 19.6, P , 0.05), Astropecten irregularis (F1,45 25.2, P , 0.01) and Pagurus spp (ANOVA: F 1,4 5 9.3, P , 0.05). The scavenger responses to mono-speci®c baits of damaged Asterias rubens and Buccinum undatum were considerably smaller than to mixed or bivalve baits. After baiting with damaged A. rubens, only ¯at®sh increased signi®cantly (ANOVA: F1,4 5 10.3, P , 0.05), and after baiting with damaged B. undatum, signi®cant increases were observed in Callionymus lyra (ANOVA: F1,45 36.4, P , 0.01), ¯at®sh (ANOVA: F 1,4 5 106.6, P , 0.01), and Liocarcinus spp (ANOVA: F1,4 5 17.2, P , 0.05). The results described in the preceding paragraphs are summarised in Table 4. Direction of arrival of Asterias rubens was signi®cantly correlated with water current direction when a number of baits were used: badly damaged Cancer pagurus (Circular± Circular Correlation Coef®cient: nr 2 5 10.02, P , 0.05), badly damaged A. rubens (CCCC: nr 2 5 11.19, P , 0.05), damaged Aequipecten opercularis (CCCC: nr 2 5 23.20, P , 0.001), and undamaged A. opercularis (CCCC: nr 2 5 31.08, P , 0.001), but was not correlated with water direction during two control periods studied (CCCC: nr 2 5 1.52, P . 0.05 and nr 2 5 9.29, P . 0.05). In all signi®cant cases, the mean direction of arrival of A. rubens was opposite to that of the prevailing water current. Surprisingly, however, signi®cant correlations with water current direction were not detected after the two baitings with mixed damaged benthos (CCCC: nr 225 5.11, P . 0.05 and nr 5 1.56, P . 0.05).

4. Discussion

The scavenging assemblages identi®ed in this study were dominated by invertebrates, mainly the star®sh, Asterias rubens, but also Astropecten irregularis, and the crabs Pagurus spp and Liocarcinus spp. In contrast to other studies where gadoid ®sh have been abundant (Kaiser and Spencer, 1996), ®sh species occurred only in low abun- dances. The sizes of the aggregations, although signi®cantly larger than background L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 125 levels, are relatively small. Aggregations of hermit crabs, Pagurus bernhardus,ofupto 330 m22 at baited bags in the Irish Sea have been recorded. These differences may be related to background population densities of the scavenging species, or to the availability of natural food sources (Ramsay et al., 1997b). The use of red light to ®lm animal activity at night enabled a more realistic assessment of the scavenger assemblage including nocturnal species, whilst minimising the arti®cial attraction of the light itself, although not completely removing it. The diel activities of both Astropecten irregularis and Cancer pagurus were signi®cantly related to the type of light used, both being more abundant under white light. C. pagurus is known to be more active at night (Bergman et al., 1990; Skajaa et al., 1998), so light avoidance behaviour may have been expected, but was not observed. Another Astropec- ten species (A. articulatus) is known to be crepuscular (Bedding®eld et al., 1991), so conceivably their observed nocturnal activity may have been related to the low light levels from the video work. In a previous experiment in the Irish Sea, scavenger numbers were recorded as peaking between 8 and 14 h after baiting (Kaiser and Spencer, 1996). However, the ¯uctuations apparent in Figs. 3 and 4 (present study) suggest that more peaks may have been observed had the period of observation been extended. In the present study, the aggregations of some species remained for up to 72 h after baiting, before returning to background levels. This may have been related to the size of the aggregation: fewer scavengers would take longer to consume a bait of a given size. In the case of Asterias rubens, the prolongation of the aggregation may be related to their post-prandial behaviour: having fed to satiation, A. rubens generally remain motionless for some time (Jangoux, 1982), presumably digesting their meal, and would continue to be recorded on the video. The change in composition of the assemblage observed after baiting (Fig. 5) implies differential reactions to the bait in the species recorded; some aggregated, some did not. One of the most pronounced differences was the increased relative abundance of Pagurus spp; it comprised 19% of the total assemblage abundance averaged over the 3 days before baiting, and 46% after. This observation concurs with other work in the Irish Sea (Ramsay et al., 1996, 1997a) and in the Clyde (Nickell and Moore, 1991, 1992a,b) where P. bernhardus was found to be the most abundant scavenger. The observed change in assemblage composition is obviously very localised and speci®c to the small area surrounding the baited camera: it is simply a temporary disruption of the spatial distribution of these species over the seabed. Some authors (e.g., Kaiser and Ramsay, 1997; Ramsay et al., 1997b; Thrush et al., 1998) have suggested that community level changes may already have occurred on heavily ®shed areas. In effect, the dietary advantages conferred upon scavenging species (e.g. Pagurus spp and Asterias rubens) shown here, may lead to long-term bene®ts at the population level. Although the present study is of too short a time span to verify this population change, the mechanism by which such a shift could occur is demonstrated. Given the ubiquity of A. rubens around the Isle of Man, this species must play a major role in discard recycling, and previous studies have suggested an increase in A. rubens abundance in the Irish Sea since the onset of the commercial scallop ®shery in the 1930s (Hill et al., 1999). 126 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

There appeared to be an elevated presence of Asterias rubens after baiting with damaged Aequipecten opercularis compared to undamaged ones, but it was not demonstrable statistically. The difference appears to lie not in the initial aggregation levels, but in the persistence of the aggregation over 3 days, implying that A. rubens are initially attracted just as strongly to undamaged queen scallops as to damaged ones, but disperse more rapidly. It is interesting to note that the undamaged A. opercularis (only 15 individuals) still attracted up to seven times the unbaited levels of A. rubens.Itis generally believed that scavengers are attracted by the odour plume released from damaged tissues (Sainte-Marie and Hargrave, 1987), although other stimuli may be involved (Lapointe and Sainte-Marie, 1992). In this case, the excretion of normal metabolic by-products from such an unnaturally high density of undamaged A. opercularis may have stimulated aggregation. It seems likely that even undamaged animals among the discards will be exposed to increased predator encounter rates and concomitant elevated stress levels (Crowder and Murawski, 1998), for example Ramsay and Kaiser (1998) showed that apparently undamaged , which had been rolled to simulate movement in ®shing gear, were signi®cantly less likely to exhibit an escape response than unrolled ones. The Irish Sea scallop ®shery is subject to various legislations, including a minimum legal landing size of 110 mm. The dredge teeth and belly ring diameter exert some size selection on the target species (Dare et al., 1993), but undersized scallops will inevitably be retained, and subsequently discarded, especially on grounds with large amounts of dead shell, which quickly block up the dredge belly. These scallops, which may be damaged or exhausted, in addition to those damaged by the dredge on the seabed but not retained, may suffer considerable predation mortality. Related scallop restocking studies have demonstrated high predation levels in small scallops seeded in unnaturally high densities (Barbeau et al., 1996, 1998). This has obvious implications for ®sheries management. Mono-speci®c baits of Asterias rubens, Buccinum undatum and Pecten maximus appear to be generally less attractive to scavengers than the mixed baits, and both the damaged and undamaged Aequipecten opercularis baits. It seems likely that A. opercularis, a component of the mixed baits, is a particularly attractive prey to scavenging species. This may be due to a combination of its energy value as a food source and its relative accessibility (the queen scallop shell is much more fragile than that of a scallop or a ). It is also possible that Echinus esculentus is an attractive prey, but mono-speci®c baits of this species were not used. The lack of trials using all by-catch taxa individually obviously limits the interpretation possible here, but it can be concluded that scavengers can discern prey species from odour plumes. This has previously been demonstrated for some scavenging species, e.g. hermit crabs (Thacker, 1996; Rittschof and Hazlett, 1997). The movement of scavengers towards dredge tracks or discarded by-catch is probably stimulated by chemosensory stimuli (Warner, 1979; Lapointe and Sainte-Marie, 1992; Nickell and Moore, 1992a,b). The bottom current velocity will therefore play an important role in both the distribution of carrion odours, and the speed of aggregation of the different scavenging species. There is some supporting evidence from these experiments, i.e. Asterias rubens move up current towards the bait. However, the lack of L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 127 correlation observed after baiting with mixed damaged benthos appears to contradict this, especially as these baits attracted the greatest number of scavengers. Possibly the larger total numbers of animals aggregating on these baits provided other cues (e.g., auditory) which attracted randomly distributed A. rubens from the surrounding area (including upstream). The ®ndings of this study are speci®c to the area under investigation, and other studies (e.g., Ramsay et al., 1997b) have identi®ed large variation between scavenger responses in different habitats. However, the present study was sited in a closed area adjacent to one of the most productive scallop grounds in the Irish Sea (Brand et al., 1991), and so responses may be considered typical for the seabed types dredged for scallops in this area. These ®ndings demonstrate that a wide range of scavenger species bene®t from the input of by-catch material, over an extended time period after discarding. The prey species vary in attractiveness to scavengers, and there is evidence to suggest that the inclusion of queen scallops, Aequipecten opercularis, in the discards may increase the size of scavenger aggregation. The damage sustained by the by-catch may not be as instrumental in determining attraction of predators as initially assumed. It is apparent that undamaged animals in unnaturally high densities may be suf®cient to bring about a scavenger aggregation.

Acknowledgements

The authors would like to thank the crew of the RV Sula and all the divers who helped, especially M. Bates and M. Mosley. This study was funded by the Ministry of Agriculture, Fisheries and Food as part of contract number CSA 2332. The work also drew on the resources provided by the Isle of Man Department of Agriculture, Fisheries and Forestry scallop research project for which we are most grateful. [RW]

References

Barbeau, M.A., Hatcher, B.G., Scheibling, R.E., Hennigar, A.W., Taylor, L.H., Risk, A.C., 1996. Dynamics of juvenile sea scallop () and their predators in bottom seeding trials in Lunenburg Bay, Nova Scotia. Can. J. Fish. Aquat. Sci. 53 (11), 2494±2512. Barbeau, M.A., Scheibling, R.E., Hatcher, B.G., 1998. Behavioural responses of predatory crabs and sea stars to varying density of juvenile sea scallops. 169 (1/2), 87±98. Batschelet, E., 1981. Circular Statistics in Biology. Academic Press, London, 371 pp. Bedding®eld, S.D., McClintock, J.B., Marion, K., Hopkins, T.S., 1991. Feeding biology of the asteroid Astropecten articulatus. Am. Zool. 31 (5), 28A. Bergman, M.J.N., Fonds, M., Hup, M., Stam, A., 1990. Direct effects of beam trawl ®shing on benthic fauna in the North Sea. ICES, Copenhagen, CM1990/mini:11, pp. 1±19. Bradshaw, C., Veale, L.O., Hill, A.S., 2000. The effect of scallop dredging on Irish Sea benthos: experiments using a closed area. In: Burnell, G. (Ed.), Coastal Shell®sh Ð A Sustainable Resource. Kluwer Academic, Dordrecht, in press. 128 L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129

Brand, A.R., Allison, E.H., Murphy, E.J., 1991. North Irish Sea scallop ®sheries: a review of changes. In: Shumway, S.E., Sandifer, P.A. (Eds.), An International Compendium of Scallop Biology and Culture. World Aquaculture Society, Baton Rouge, pp. 204±218. Britton, J.C., Morton, B., 1994. Marine carrion and scavengers. Oceanogr. Mar. Biol. Annu. Rev. 32, 369±434. Caddy, J.F., 1973. Underwater observations on the tracks of dredges and trawls and some effects of dredging on a scallop ground. J. Fish. Res. Bd Can. 30 (2), 173±180. Chapman, C.J., Mason, J., Kinnear, J.A.M., 1977. Diving observations on the ef®ciency of dredges used in the Scottish ®shery for the scallop, Pecten maximus (L.). Scottish Fisheries Research Report, No. 10. Marine Laboratory, Aberdeen (16 pp.). Clarke, K.R., Warwick, R.M., 1994. Change in Marine Communities: An Approach to Statistical Analysis and Interpretation. Natural Environmental Research Council, Swindon, 144 pp. Crowder, L.B., Murawski, S.A., 1998. Fisheries bycatch: implications for management. Fisheries 23 (6), 8±17. Dare, P.J., Key, D., Darby, C.D., Connor, P.M., 1993. The ef®ciency of spring-loaded dredges used in the western English Channel ®shery for scallops, Pecten maximus (L.). Ministry of Agriculture Fisheries and Food, Directorate of Fisheries Report. Dayton, P.K., Thrush, S.F., Tundi Agardy, M., Hofman, R.J., 1995. Environmental effects of marine ®shing. Aquat. Conserv. 5, 205±232. Furness, R.W., 1996. A review of seabird responses to natural or ®sheries-induced changes in food supply. In: Greenstreet, S.P.R., Tasker, M.L. (Eds.), Aquatic Predators and Their Prey. Blackwell Scienti®c, Oxford, pp. 166±173. Hill, A.S., Veale, L.O., Pennington, D., Whyte, S.G., Brand, A.R., Hartnoll, R.G., 1999. Changes in Irish Sea benthos: possible effects of 40 years of dredging. Estuarine Coastal Shelf Sci. 48 (6), 739±750. Jangoux, M., 1982. Food and feeding mechanisms: asteroidea. In: Jangoux, M., Lawrence, J.M. (Eds.), Nutrition. A.A. Balkena, Rotterdam, pp. 117±159. Jennings, S., Kaiser, M.J., 1998. The effects of ®shing on marine ecosystems. Adv. Mar. Biol. 34, 201±351. Kaiser, M.J., Ramsay, K., 1997. Opportunistic feeding by dabs within areas of trawl disturbance: possible implications for increased survival. Mar. Ecol. Prog. Ser. 152, 307±310. Kaiser, M.J., Spencer, B.E., 1994a. Fish scavenging behaviour in recently trawled areas. Mar. Ecol. Prog. Ser. 112, 41±49. Kaiser, M.J., Spencer, B.E., 1994b. A preliminary assessment of the effect of beam on a benthic community in the Irish Sea. In: de Groot, S.J., Lindeboom, H.J. (Eds.), Environmental Impact of Bottom Gears On Benthic Fauna in Relation To Natural Resources Management and Protection of the North Sea. Netherlands Institute for Sea Research, Texel, pp. 87±94. Kaiser, M.J., Spencer, B.E., 1996. Behavioural responses of scavengers to beam trawl disturbance. In: Greenstreet, S.P.R., Tasker, M.L. (Eds.), Aquatic Predators and Their Prey. Blackwell Scienti®c, Oxford, pp. 116±123. Lapointe, V., Sainte-Marie, B., 1992. Currents, predators, and the aggregation of the gastropod Buccinum undatum around bait. Mar. Ecol. Prog. Ser. 85 (3), 245±257. Medcof, J.C., Bourne, N., 1964. Causes of mortality of the sea scallop, Placopecten magellanicus. Proc. Natl. Shell®sh Assoc. 53, 33±50. Murawski, S.A., Serchuk, F.M., 1989. Environmental Effects of Offshore Dredge Fisheries For Bivalves. ICES, Copenhagen, CM1989/K:27, 12 pp. Nickell, T.D., Moore, P.G., 1991. The behavioural ecology of epibenthic scavenging invertebrates in the Clyde Sea area: ®eld sampling using baited traps. Cah. Biol. Mar. 32 (3), 353±370. Nickell, T.D., Moore, P.G., 1992a. The behavioural ecology of epibenthic scavenging invertebrates in the Clyde Sea area: laboratory experiments on attractions to bait in static water. J. Exp. Mar. Biol. Ecol. 156 (2), 217±224. Nickell, T.D., Moore, P.G., 1992b. The behavioural ecology of epibenthic scavenging invertebrates in the Clyde Sea area: laboratory experiments on attractions to bait in moving water, underwater TV observations in situ and general conclusions. J. Exp. Mar. Biol. Ecol. 159 (1), 15±35. Polis, G.A., Holt, R.D., Menge, B.A., Winemiller, K.O., 1996. Time, space and life history: in¯uences on food webs. In: Polis, G.A., Winemiller, K.O. (Eds.), Food Webs: Integration of Pattern and Dynamics. Chapman & Hall, London, pp. 438±447. L.O. Veale et al. / J. Exp. Mar. Biol. Ecol. 255 (2000) 111 ±129 129

Ramsay, K., Kaiser, M., Hughes, R., 1996. Changes in hermit crab feeding patterns in response to trawling disturbance. Mar. Ecol. Prog. Ser. 144, 63±72. Ramsay, K., Kaiser, M.J., 1998. Demersal ®shing disturbance increases predation risk for whelks (Buccinum undatum L.). J. Sea Res. 39 (3/4), 299±304. Ramsay, K., Kaiser, M.J., Hughes, R.N., 1997a. A ®eld study of intraspeci®c competition for food in hermit crabs (Pagurus bernhardus). Estuarine Coastal Shelf Sci. 44, 213±220. Ramsay, K., Kaiser, M.J., Moore, P.G., Hughes, R.N., 1997b. Consumption of ®sheries discards by benthic scavengers: utilization of energy subsidies in different marine habitats. J. Anim. Ecol. 66, 884±896. Rittschof, D., Hazlett, B.A., 1997. Behavioural responses of hermit crabs to shell cues, predator haemolymph and body odour. J. Mar. Biol. Assoc. UK 77 (3), 737±751. Sainte-Marie, B., Hargrave, B.T., 1987. Estimation of scavenger abundance and distance of attraction to bait. Mar. Biol. 94, 431±443. Skajaa, K., Ferno, A., Lokkeborg, S., Haugland, E.K., 1998. Basic movement pattern and chemo-oriented search towards baited pots in edible crab (Cancer pagurus L.). Hydrobiologia 372, 143±153. Thacker, R.W., 1996. Food choices of land hermit crabs (Coenobita compressus H. Milne Edwards) depend on past experience. J. Exp. Mar. Biol. Ecol. 199 (2), 179±191. Thrush, S.F., Hewitt, J.E., Cummings, V.J., Dayton, P.K., 1995. The impact of habitat disturbance by scallop dredging on marine benthic communities: what can be predicted from the results of experiments? Mar. Ecol. Prog. Ser. 129, 141±150. Thrush, S.F., Hewitt, J.E., Cummings, V.J., Dayton, P.K., Cryer, M., Turner, S.J., Funnell, G.A., Budd, R.G., Milburn, C.J., Wilkinson, M.R., 1998. Disturbance of the marine benthic habitat by commercial ®shing: impacts at the scale of the ®shery. Ecol. App. 8 (3), 866±879. Veale, L.O., Hill, A.S., Hawkins, S.J., Brand, A.R., 2000. Effects of long-term physical disturbance by commercial scallop ®shing on subtidal epifaunal assemblages and habitats. Mar. Biol. 137, 325±337. Warner, G.F., 1979. Aggregation in . In: Larwood, G., Rosen, B.R. (Eds.), Biology and Systematics of Colonial Organisms. Academic Press, London, pp. 375±396.