- Email: [email protected]
Recovery of sediments after cessation of marine fish farm production
Aquaculture 235 (2004) 315 – 330 www.elsevier.com/locate/aqua-online
Recovery of sediments after cessation of marine fish farm production Paula M.F. Pereira a,1, Kenneth D. Black a,*, Donald S. McLusky b, Thom D. Nickell a a
Dunstaffnage Marine Laboratory, Scottish Association for Marine Science, Oban, Argyll, Scotland PA37 1QA, UK b Department of Biological and Molecular Sciences, Stirling University, Stirling, Scotland FK9 4LA, UK Received 8 May 2003; received in revised form 11 December 2003; accepted 15 December 2003
Abstract Following cessation of fish production at a fish farm site in Loch Creran, Scotland, a study of the recovery of the benthic environment was undertaken. Sediment samples for macrofauna and geochemical parameters (redox potential, organic carbon, oxygen flux) were collected over a period of 15 months from three stations following a gradient of impact from the former fish farm site. The data collected were analysed by a combination of uni-and multivariate statistical methods. The macrobenthic community at the two stations furthest from the fish cage site showed signs of recovery with time in terms of indicator species, number of species and abundance, being, however, still moderately to slightly disturbed at the end of this study. At the station nearest to the former fish cage site, recovery of the macrobenthic community was also evident, but this station was still highly impacted 15 months after fish production ceased, with opportunistic species dominant. Fifteen months after fallowing, highly reduced conditions were still persistent in subsurface sediments at the stations on the periphery of the former fish cage site. Bulk sediment organic carbon, although an indicator of a spatial gradient, was not found to be a significant indicator of recovery. Combinations of different environmental parameters appear to affect different stages of benthic recovery with sediment oxygen uptake as the main observed parameter conditioning early stages of macrobenthic succession. D 2004 Elsevier B.V. All rights reserved. Keywords: Recovery; Oxygen Flux; Macrofauna; Sediment redox; Fish farm; Benthos
* Corresponding author. E-mail address: [email protected] (K.D. Black). 1 Present address: IPIMAR, Av. 5 de Outubro, 8700 Olhao, Portugal. 0044-8486/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.aquaculture.2003.12.023
316
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
1. Introduction Intensive salmon culture in sea cages often results in changes to the benthos through enrichment of sediments with organic wastes products (reviewed by Pearson and Black, 2001). Fallowing in marine fish culture involves the removal of all fish and nets from sea cage sites for different periods of time. When using this technique for disease control, production is usually interrupted for only a few months (Bron et al., 1993; FRS, 2001). Fallowing can also be used to reduce the build up of sediments beneath the cages and thus minimise any potential risk of ‘self-pollution’ (Beveridge, 1987; Lumb, 1989). This is achieved either by repositioning the cages or by alternately farming at two separate sites, usually for 12 to 24 months (one growing cycle). In Scotland, the frequency of rotation and the length of the fallowing period are at the operator’s discretion and varies considerably depending on company policy and production pressure. Regulators may also stipulate long fallow periods to allow sediments to recover where the input has been beyond the local assimilative capacity of the environment and this may allow farming to continue at an economic scale for alternate cycles. This approach is taken to ensure that sediment quality standards are met. The biogeochemical and biological processes that drive recovery are, however, poorly studied and the appropriate fallowing time is not known for any site. Based on studies of pulp mill waste, Pearson and Rosenberg (1978) suggested that changes in macrofaunal structure are similar across both spatial and temporal gradients. Despite numerous studies having being carried out to describe spatial changes of benthic community and sediment chemical conditions in response to marine fish farming (Brown et al., 1987; Ritz et al., 1989, Weston, 1990; Ye et al., 1991; Findlay et al., 1995; Karakassis et al., 1999), relatively little information is available on the recovery from those changes. Gowen et al. (1988) and Lumb (1989), studying the effects of salmon fish farming in Scottish sea loch sediments, found that no significant benthic macrofaunal recolonisation was apparent 1 year after fish production ceased. Ritz et al. (1989), however, observed a 7-week period for recovery from moderately disturbed to undisturbed sediment conditions. Slow recovery of the macrobenthic community was reported after the removal of a salmon farm in Norway, with the community still highly dominated by Capitella capitata 1 year later (Johannessen et al., 1994). In Mediterranean waters, Karakassis et al. (1999) also observed that a fish farm site, having been stocked for 6 years, had not entirely recovered 23 months after fish cages had been removed. In Canada, Pohle et al. (2001) have also found no recovery of benthic macrofauna community 1 year after fish farming ceased. Recovery studies with other sources of organic enrichment which show similar spatial gradients as fish farm wastes, such as domestic sludge waste (Eleftheriou et al., 1982; Moore and Rodger, 1991), pulp mill waste (Rosenberg, 1972, 1973, 1976; Christie and Green, 1982) and oil spills (Cabioch et al., 1982; Gray, 1982), show biological parameters (diversity, abundance and number of species) to be similar to background after 3 to 12 or more years depending on the degree of contamination and on hydrographic parameters. The aims of the present study were: (a) to describe the effect and time scale of the cessation of the input of fish farm waste on macrobenthic succession; and (b) to evaluate which biogeochemical parameters best correlated with recovery of the benthos.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
317
2. Materials and methods 2.1. Study site The sampling site was located in the vicinity of a salmon farm in Loch Creran, a fjordic sea loch on the west coast of Scotland north of Oban (Fig. 1). The loch is 12.8km long and has two main basins separated by shallow sills. The larger outer basin, in which the sampling site was located, is 9-km long and of 49-m maximum depth. The loch is sheltered from severe wave action and has a mean freshwater input of 286 106 m3 year 1 (Edwards and Sharples, 1986). The fish farm was composed of 10 cages moored together. Each cage was a square structure (25 25 m) from which a net was hung to a depth of 11 m. The site had been used for fish farming for several years with an estimated input of 169 tonnes of organic carbon into the environment (as waste food and faeces) for the 2 years previous to July 1994 when fallowing started. The site resumed production in May 1996. Near-bottom seawater temperatures at the sampling site ranged over the study period from 6 to 14 jC, and the direction of the residual current was NW ranging from 0 to 37 cm s 1 with a mean current of 3.8 cm s 1 at 2.5 m above the sediment surface (Pereira, 1997). The seabed at the three stations was silty with median grain diameter of 0.05 and a silt-clay content of 55% (Pereira, 1997). In order to select sampling stations, preliminary transects were set in July 1994 after fish farming had ceased but while the fish cages were still in place. The degree of organic enrichment was assessed by measuring redox potential profiles (Pearson and
Fig. 1. Loch Creran with inset showing the sampling site, bathymetry and the locations of stations A, B and C positions relative to the former fish cage group.
318
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Stanley, 1979). In situ observation of the sediment by divers was also taken into consideration (presence/absence of Beggiatoa spp. mats). Three stations were selected, all at 19-m water depth and following a gradient of impact from the former cage group location with station A (no Beggiatoa spp. mats; Eh (2-cm depth) = 22 mV; Eh (4-cm depth) = 30 mV) at 33 m south east from the cage, station B (Beggiatoa spp. mats present; Eh (2-cm depth) = 153 mV; Eh (4-cm depth) = 190 mV) at 3 m from the south east edge of the cage and station C (Beggiatoa spp. mats present; Eh (2-cm depth) = 191 mV; Eh (4-cm depth) = 213 mV) at the south edge of the cage, 25 m west from station B (Fig. 1). The stations were located up-current from the cage group to minimise the distance between stations, thus allowing maximum bottom time for diving and also minimising inter-station differences in depth and substratum. Due to the presence in Loch Creran of other possible inputs of high organic enrichment such as other fish farms, a fish processing plant and a kelp processing plant, it was impossible to find a reference site with the same hydrographic and sedimentary characteristics away from the influence of the fish farm. 2.2. Sampling procedures Samples were collected 1 (4/8/1994), 2 (5/9/1994), 8 (6/3/1995), 14 (21/8/1995) and 15 (11/10/1995) months after fallowing at stations A and B. Station C was sampled at all the above dates excepting 4/8/1994 when, due to adverse weather conditions, the dive was aborted. Sediment samples were collected by SCUBA divers using core tubes (57-mm i.d. 230-mm height). This sampling technique was preferred to the use of a grab due to the precise station fixing allowed by diver coring along a fixed transect, necessary in areas of steep organic enrichment gradients. At each station and each date, three replicate sediment cores were collected for analyses of benthic macrofauna. The samples were fixed in 10% formalin buffered with borax, until further analysis. The samples were washed through a 500-Am sieve and all the retained benthos hand-sorted and identified to the lowest practical taxonomic level. Species diversity was calculated by means of the (log2) Shannon –Wiener index (Krebs, 1972). The percentages of organic carbon and nitrogen were determined from duplicate cores in set intervals (0– 1; 1 –3; 3 –6; 6 –9cm). The organic carbon and nitrogen sediment samples were measured using a Leco CHN-900 Analyser following vapour-phase acidification (Hedges and Stern, 1984). Oxidation– reduction potential (Eh) was determined in duplicate cores on the overlying water 1 cm above the sediment, and from the sediment surface at 0.5-cm intervals to 7-cm depth as described by Pearson and Stanley (1979). Results are reported from the 2- and 4-cm sediment depths only, as these are considered to be the most representative of the overall levels down the sediment column (Pearson and Stanley, 1979). Sediment metabolism was measured as oxygen consumption in undisturbed sediment cores. At each sampling date, two sediment cores from each station, with their top bung removed, were stored overnight in a container with aerated water from the sampling site at in situ temperature, in order to reduce disturbance as a result of sampling and subsequent transport. The cores were then sealed with a top which incorporated an electronic magnetic stirrer to maintain a continuous water circulation at a rate just below resuspension limit, and allowing
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
319
repeated sampling of the overlying water (Parkes and Buckingham, 1986). Oxygen flux was determined from the concentration difference between initial and final samples during incubation periods of 3 h. Dissolved oxygen was determined using the micro Winkler back-titration method (Grasshoff, 1983). A Malvern Mastersizer/E and relevant software (NEC PC 486SX running Mastersizer/ E version 1.2a) were used to determine the median grain size (Md), sediment percentage silt-clay ( < 63 Am) and sediment sorting (rI) from core samples collected from each station at each date. Sea water temperature was measured with a digital diving depth meter in the water overlying the sediment. 2.3. Statistical analyses Multivariate analyses followed methods described by Clarke (1993) using the PRIMER (Plymouth Routines in Multivariate Ecological Research) software package. To satisfy statistical assumptions for significance-testing procedures, all environmental data were log10-transformed and faunal abundance data square-root transformed. The CLUSTER method was used to compute a similarity matrix based on the Bray– Curtis similarity coefficient. The similarity matrix was subjected to hierarchical, agglomerative classification employing group-average sorting to classify the stations based on faunal groupings. Non-metric Multidimensional Scaling ordination analyses (MDS) were performed on the similarity matrices obtained using the Bray– Curtis similarity coefficient. To test for statistically significant differences between the macrobenthos structures at the different stations and sampling dates, a two-way crossed ANOSIM test was used. The BIO-ENV method was used in order to demonstrate the most important environmental variables related to faunal patterns.
3. Results An extensive display of the chemical and physical results and the list of species per each core sampled with respective abundance are given in Pereira (1997). 3.1. Physical and chemical data The organic carbon content of the sediment surface layer on the edge of the cage group (station C) was about twice of those values found 55 m away (station A) (Fig 2). The sediment oxygen consumption was also elevated (Fig. 3) and the sediment redox potential was substantially depressed (Fig. 4). At the two first sampling dates white mats of the sulphur-oxidizing bacteria, Beggiatoa spp., covered the seabed at station C (Table 1). At station C the increase in organic carbon from March to August 1995 (months 8 to 14) at all depth horizons is most accentuated at the 1 –3-cm depth horizon (Fig. 2). In month 15, the organic carbon decreased in the upper sediment layers (0 –1 and 1– 3 cm), but increased in the deeper sediment layer (3– 6 cm). At station C, pockets of Beggiatoa spp. reappeared 14 and 15 months after fallowing (August and October 1995), indicative of hypoxic sediment conditions.
320
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 2. Temporal changes in organic carbon percentage in the sediment, at different depth horizons at the three stations sampled.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 3. Temporal changes in sediment oxygen flux rate (mmol d
1
m
321
2
) at the three stations sampled.
Fig. 4. Temporal changes in redox potential values (mV) at 2- and 4-cm sediment depth at the three stations sampled.
322
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Table 1 Average relative abundance (%) per station (A, B, C) and different fallowing times of the most abundant polychaete species Station
A
Months of fallowing 1 Presence of Beggiatoa spp. C.spp (ind m 2) 6 M.frag (ind m 2) 9 M.ful (ind m 2) 0 P.fallax (ind m 2) 4 39* O.hart (ind m 2) S.infl (ind m 2) 2
B 2
8
14
15
1
C 2
8
14 15 + + 9 0 0 0 26* 51* 43* 5 3 16 24* 34* 29* 14 1 5 28* 21* 0 0 0 0 16 17 0 0 1 2 11 3 13 0 0 0 15 20 41* 0 0 0 21 20 4 0 0 1 6 12 4 0 0 4 4 0
1 2 8 14 15 + + + + – 74* 56* 28* 10 – 0 3 25 14* – 17 0 0 8 – 0 0 3 1 – 1 0 0 0 – 0 2 0 0
(*) Dominant polychaete species. C. spp: Capitela spp.; M. frag: Mediomastus fragilis; M. ful: Malacocerus fuliginosus; P. fallax: Prionospio fallax; O. hart: Ophryotrocha hartmanni; S. infl: Scalibregma inflatum.
3.2. Macrofaunal data A total of 71 taxa was identified, with a minimum of 9 taxa and a maximum of 44 taxa found per station at all dates (Fig. 5). The number of species increased with distance from the fish farm site and in general with fallowing time. While station A showed a seasonal trend in macrofaunal abundance, station B and C abundance generally decreased with fallowing time (Fig. 5). At the beginning of the study, Ophryotrocha hartmanni and Mediomastus fragilis were the most abundant species present at station A (Table 1). Capitella sp. was the third most abundant polychaete accounting for 6– 9% of the total abundance of that station. Malacocerus fuliginosus was never present in station A samples. M. fragilis and Prionospio fallax accounted for 42% of the total abundance 15 months after fallowing at station A. Capitella sp., M. fuliginosus and O. hartmanni were the dominant species (64% and 88% at Station B and 91% at Station C) at Stations B and C at the two first sampling dates. Fifteen months after fallowing, M. fragilis and P. fallax were the most abundant polychaete species at station B. M. fragilis was found in smaller numbers at station C than at the other two stations but showed a steady increase over time and was the annelid with the highest density 15 months after fallowing, closely followed by Capitella sp. and M. fuliginosus (Table 1). The Shannon – Wiener diversity index at station A fluctuated between 3.1 and 3.7 (Fig. 6). The species diversity in station B increased until it reached 3.6, 14 months after fallowing started, decreasing to 3.0 at 15 months. The species diversity at station C, although increasing with fallowing time, was consistently lower than 3.0 throughout the survey. 3.3. Multivariate statistics The multidimensional scaling configuration (MDS) for the abundance data (Fig. 7) shows that the three stations follow a similar nonlinear temporal trajectory. The rate of change was different between stations, with station B macrofaunal structure changing
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 5. Temporal changes in total number of taxa (in 0.0075 m2) and mean abundance (ind m stations sampled.
323
2
) at the three
Fig. 6. Temporal changes in species diversity (log2 Shannon – Wiener index) at the three stations sampled.
324
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 7. MDS configuration of square-root transformed species-abundance mean data at each station (A, B, C) and at each sampling data (1, 2, 8, 14 and 15 months of fallowing).
(recovering) at a higher rate than station C. Although there was a similar tendency for all stations to move towards recovery, all stations were found to be significantly different from each other at the 5% level (Table 2). Again, all pairs of dates were found to be significantly different with dissimilarity generally decreasing in time (Table 2). The BIO-ENV multivariate showed the combination of particle diameter, oxygen flux and organic carbon variables as the match that best grouped the stations in a manner consistent with the faunal pattern. The match, however, was not very successful (0.57) Table 2 Two-way crossed ANOSIM test for differences between (a) stations and (b) sampling dates, using square-root transformed species abundance data, and the Bray – Curtis similarity measure (three replicates at each station) (a) Tests for differences between stations groups Sample statistic (Global R): 0.770 Significance level of sample statistic: 0.001% Groups used
Statistical value R
Significance level (%)
(A, B) (A, C) (B, C)
0.815 0.991 0.657
0.001 0.001 0.001
(b) Tests for differences between time groups (as months of following) Sample statistic (Global R): 0.806 Significance level of sample statistic: 0.001% Groups used
Statistical value R
Significance level (%)
2, 8 2, 14 2, 15 8, 14 8, 15 14, 15
0.938 1.000 1.000 0.617 0.728 0.519
0.001 0.001 0.001 0.001 0.001 0.001
The samples from stations A and B in August 1994 were not analysed as there were no comparative data from station C.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
325
Table 3 Combinations of the 7 environmental variables (organic carbon, oxygen flux, redox potential, particle diameter, silt-clay percentage, sediment sorting, sea water temperature) yielding the best match of biotic and abiotic similarity matrices, as measured by weighted Spearman rank correlation qN; correlation coefficient in brackets BIO-ENV; weighted Spearman rank correlation
Best variable combination
Station A
Station B
Station C
O2, OC (0.66)
O2 (0.81)
O2 (0.94)
O2—oxygen flux; OC—organic carbon.
indicating that the environmental variables measured only partially explain the distribution of the faunal data. Since the stations sampled reflect different influences from the fish farm waste (due to the distance from the source), it was thought that the weight of the different environmental variables ‘explaining’ the community structure at the different stations might also be distinct. Table 3 shows the single and combination of environmental variables that best group the temporal variation of individual stations in a manner consistent with the faunal pattern. The correlation coefficient decreases with distance from the fish farm site. No physical variable (sea water temperature, sediment particle size, sorting coefficient, sediment clay fraction) was found to be included in the best match. Oxygen consumption was found to be the environmental parameter that better correlated with the faunal variation in time at stations B and C. At station A the combination of oxygen and concentration of organic carbon in the sediment showed the best correlation with the faunal pattern.
4. Discussion The degree and types of disturbance observed at the fish farm site in Loch Creran were broadly consistent with other studies on effects of fish farm waste on sediment chemical and biological properties (Brown et al., 1987; Kaspar et al., 1988; Ritz et al., 1989; Weston, 1990, Holmer and Kristensen, 1992; Karakassis et al., 1999). Although seasonal variations in ‘natural’ organic material input, sedimentation, temperature and larval availability will have an additional effect on the environmental and biological parameters measured, the general worsening of sediment and biological conditions with proximity to the fish farm site, and the gradual improvement with fallowing time, suggests that the effects found were mainly related to the cessation of input of fish farm waste. The consumption of oxygen by chemical or biological processes could not be distinguished using the data collected. The high variability between replicate measurements of oxygen uptake could reflect spatial patchiness in sedimentation of faecal waste and food pellets and also the patchy effects of epi-faunal activities. The initial oxygen uptake values of 80– 100 mmol m 2 d 1 were typical of enriched sediments (Holby and Hall, 1991; Hargrave et al., 1993; Phillips, 1995). Much higher oxygen flux rates (435 mmol m 2 day 1;Nickell et al., 2003) have subsequently been measured under a different farm in Loch Creran, but this was during a period of high production rather than ca. 1 month after production ceased as in the first samples in the present study. In March 1995, 8 months after fallowing, oxygen uptake rates (30 –50 mmol m 2 d 1) were still greater than estimates for sea-lochs not subjected to fish farming (5 – 25 mmol
326
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
m 2 d 1) (Davies, 1975; Parkes and Buckingham, 1986; Hall et al., 1990; Nickell et al., 2003). Organic material available to the benthos can be divided into two decomposable fractions of considerably different reactivity plus a non-metabolizable fraction (Westrich and Berner, 1984; Grant and Hargrave, 1987). The initial high oxygen consumption is likely to be due to the residual high levels of highly degradable organic carbon originating from the fish farm. The percentage of organic carbon in the surface sediment (0 – 1 cm) observed throughout the survey at stations B and C (2– 4%) was within the range of background levels found in other sea-lochs (Rigdway and Price, 1987; Nickell et al., 1995) and similar to values observed for Loch Creran (2 –5%) before the introduction of fish farms (Ansell, 1974) and to previous years’ background values when the fish farm was active (SEAS, 1991; SEPA, 1992). However, the percentage of organic carbon in the sediment (0– 3 cm) decreased with distance from the fish farm site, being two to three times higher at the edges of the cages (stations B and C) than 30 m away (station A); similar results being found at other fish farm sites (Brown et al., 1987; Hall et al., 1990; Weston, 1990; Ye et al., 1991; Holmer and Kristensen, 1992; Karakassis et al., 1999). The decrease of organic carbon in the surface sediment at stations B and C after fish farm production ceased, approaching the values at station A, is also an indication that the fish farm waste was the main origin of the organic carbon present in the sediment at the beginning of the study. The presence of elevated organic carbon, relative to station A, at stations B and C to a depth of 6 cm, 15 months into fallowing, indicates the persistence of contamination by fish farm waste in these subsurface sediments, consistent with low redox potential values found at the same depths. Moore and Rodger (1991) noted that although the first layers of sediment showed signs of recovery, deeper sediments remained contaminated for longer. This could affect the degree and time of response to further disturbances, with natural disturbances having a greater effect than in non depth-contaminated sediments. Seasonal variations in sediment redox potential at stations B and C followed seasonal temperature changes, indicating an increased sediment metabolism during the summer months of maximum temperature but the variables were not significantly correlated as has been found in previous fish farm studies (Holmer and Kristensen, 1992; Hargrave et al., 1993). Most biological univariate indices used illustrated gradual faunal changes over time, from highly disturbed (stations B and C) to moderately disturbed conditions. The seasonal influence, however, appears to have a strong effect over recovery with all these parameters regressing over the summer months, 14 months into fallowing. This effect is represented by the nonlinear trajectory of the MDS distribution of the individual stations with time (Fig. 7) where initial recovery towards the upper left is succeeded by a regression downwards for all stations (and subsequently to the right for stations A and C at the final sampling). Karakassis et al. (1999) related the regression of recovery observed in their study to the increase of degrading organic material buried after the peak of benthic algal production due to pulses of phosphorous released after intense fluctuations between reduced and oxidised conditions in the sediment after cessation of fish farming. Scottish sea lochs are typically turbid and benthic algal production is insignificant at the depths of the stations studied here; therefore, the
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
327
similar trends observed in the present study must be explained by fresh inputs of natural organic material (river runoff and phytoplankton production) and increased summer water temperature rather than stimulated in situ production. Varying degrees of recovery were observed at the stations studied, indicating that the speed of recovery decreases with increasing fallowing time. Although stations B and C were both highly disturbed at the beginning of the study as indicated by the high dominance by opportunistic species, the times between the successive stages of recovery were different for both stations, with station B recovering more rapidly than station C as indicated by the species succession and univariate and multivariate results. Fifteen months after the fish farm loading ceased, the highly disturbed benthic structure at station C, although improving, was still dominated by opportunistic species characteristic of highly disturbed sediments, such as Capitella sp, M. fragilis and M. fuliginosus. After 15 months of fallowing, the benthic structure at station B was characteristic of a moderately disturbed community. Station A, with a benthic structure characteristic of moderately disturbed sediments at the beginning of the study, still showed indications of being disturbed 15 months after fallowing. Lumb (1989) and Gowen et al. (1988), studying the effects of fish farms waste on Scottish sea loch sediments, found results similar to station C, with the sediment still organically enriched and no significant macrofaunal recolonisation appearing to have occurred 1 year after fish production had ceased. Slow recovery of the macrobenthic community was also reported after the removal of a salmon farm in Norway, with the community still highly dominated by C. capitata 1 year later (Johannessen et al., 1994). In Canada, Pohle et al. (2001) found no recovery of the benthic macrofauna community 1 year after fish farming ceased. Hargrave et al. (1993) compared methods for detecting organic impacts near salmon farms on the east coast of Canada and showed that total sulphide, benthic O2 and CO2 exchange, and redox potential were the most sensitive measures of 20 variables tested. When excluding site specific properties by analysing temporal variations at the individual stations, the correlation coefficient between oxygen uptake and macrobenthic fauna increased with proximity to the waste source, with this parameter appearing to be the main factor associated with macrobenthic changes at the two stations near to the fish farm cages (stations B and C). The decrease in correlation coefficient with distance from the fish farm site, and the fact that different groups of environmental parameters were found to correlate with faunal distribution at the different stations, indicates that the significance of individual environmental parameters varies with the degree of disturbance. This is likely to be a complex feedback between the microbial degradation of organic matter and sediment colonisation by tolerant, bioturbating macroorganisms that act to increase the flux of electron acceptors in the sediment while themselves contributing to the overall sediment oxygen demand. Although the amount of organic matter is believed to be the main factor influencing macrofaunal community structure after deposition of organic enriched waste (Pearson and Rosenberg, 1978; Moore and Rodger, 1991), the percentage of organic carbon in the sediment was not found in this study to be a good indicator of temporal changes in the macrobenthic community. This is in accordance with other author’s results (Dauer and Conner, 1980; Ye et al., 1991; SEPA, 1992) and it may be a result, in part, of the fact that the correlation between organic carbon and community changes depends on the labile fraction of the organic
328
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
matter (Grant and Hargrave, 1987) and not on the bulk sediment organic carbon measured in the present and previous studies, and also, in part, on the burial processes. Although at station A most sediment chemical variables such as redox potential and organic carbon showed levels characteristic of unenriched sediments, the macrobenthos structure indicated a moderate disturbance. In a recovery study from sludge waste, Eleftheriou et al. (1982) also found that 8 months after sludge addition ceased, the sediment (organic carbon and redox potential) had returned to control levels but the fauna (diversity, abundance and species numbers) had not. The same effect was observed on a spatial scale in active fish farms (Brown et al., 1987; Weston, 1990; SEPA, 1992) with effects of fish farm waste on the benthic community still apparent even when sediment chemistry (redox potential, organic carbon, water-soluble sulphide) was similar to background levels. On a spatial scale, this effect indicates that the fauna are sensitive to enrichment at levels not detectable with the gross chemical measures used. On a temporal scale, however, we also need to take into consideration that the recovery of the fauna should be slower than the recovery of the chemical conditions since the recruitment of less pollution-tolerant species can only take place after reestablishment of the environmental conditions. In Scotland, fallowing of fish farm sites is recommended by the regulator for the purpose of disease control (SOAEFD, 1996), but using this technique to avoid the build up of organic wastes in sediments is more controversial. Improvement of management techniques and diets minimises waste and increases food efficiency, so that less carbon, nitrogen and other nutrients reach the sediment beneath fish cages. The present study, however, suggests that a 2-year rotation period is unlikely to lead to the complete recovery of the most affected sediment and its associated benthic community to predisturbance conditions. Although some recovery of the benthic community takes place during fallowing, contamination in subsurface sediments persists, suggesting a rapid return to pre-fallowing conditions as soon as production resumes. For a company to be able to fallow a site while maintaining fish production, new sites would need to be granted thus increasing the total seabed surface affected by fish farming.
Acknowledgements This research was supported by the Portuguese Government, Junta Nacional de Investigacß a˜o Cientı´fica e Tecnolo´gica, Sub-Programa Cieˆncia e Tecnologia do 2j Quadro Comunita´rio de Apoio. We are grateful for the assistance provided by the staff of Scottish Seafarms and in particular by Sally Davies, by the crews of the SAMS research vessels and by the SAMS diving team especially Paul Provost.
References Ansell, A.D., 1974. Sedimentation of organic detritus in Lochs Etive and Creran, Argyll, Scotland. Mar. Biol. 27, 263 – 273.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
329
Beveridge, M.C.M., 1987. Cage Aquaculture. Fishing News Books, Farnham, Surrey. Bron, J.E., Sommerville, C., Wootten, R., Rae, G.H., 1993. Following of marine Atlantic salmon, Salmo salar L., farms as a method for the control of sea lice, Lepeophtheirus salmonis (Kroyer, 1837). J. Fish Dis. 16, 487 – 493. Brown, J.R., Gowen, R.J., McLusky, D.S., 1987. The effect of salmon farming on the benthos of a Scottish sea loch. J. Exp. Mar. Biol. Ecol. 109, 39 – 51. Cabioch, L., Dauvin, J.C., Retiere, C., Rivain, V., Archambault, D., 1982. Evolution de peuplements benthiques des fonds sedimentaires de la region de Roscoff, perturbes par les hydrocarbures de L’Amoco Cadiz. Neth. J. Sea Res. 16, 491 – 501. Christie, H., Green, N.W., 1982. Changes in the sublittoral hard bottom benthos after a large reduction in pulp mill waste to Iddefjord, Norway, Sweden. Neth. J. Sea Res. 16, 474 – 482. Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18, 117 – 143. Dauer, D.M., Conner, W.G., 1980. Effects of moderate sewage input on benthic polychaete populations. Estuar. Coast. Mar. Sci. 10, 335 – 346. Davies, I.M., 1975. Energy flow through the benthos in a Scottish sea loch. Mar. Biol. 31, 353 – 362. Edwards, A., Sharples, F., 1986. Scottish Sea Lochs: A Catalogue. Scottish Marine Biological Association, Oban. 210 pp. Eleftheriou, A., Moore, D.C., Basford, D.J., Robertson, M.R., 1982. Underwater experiments on the effects of sewage sludge on a marine ecosystem. Neth. J. Sea Res. 16, 465 – 473. Findlay, R.H., Watling, L., Mayer, L.M., 1995. Environmental impact of salmon net-pen culture on Maine marine benthic communities: a case study. Estuaries 18, 145 – 179. FRS, 2001. Scottish fish farms, annual production survey. The Scottish Executive Environment and Rural Affairs Department, Aberdeen Marine Laboratory. 47 pp. Gowen, R.J., Brown, J.R., Bradbury, N.B., McLusky, D.S., 1988. Investigations into benthic enrichment, hypernutrification and eutrophication associated with mariculture in Scottish coastal waters (1984 – 1988). Department of Biological Sciences, University of Stirling. 289 pp. Grant, J., Hargrave, B.T., 1987. Benthic metabolism and the quality of sediment organic carbon. Biol. Oceanogr. 4, 243 – 264. Grasshoff, K., 1983. Methods of Seawater Analysis, 2nd ed. Verlag Chemie, Weinheim. 419 pp. Gray, J.S., 1982. Effect of pollutants on marine ecosystems. Neth. J. Sea Res. 16, 424 – 443. Hall, P.O., Anderson, L.G., Holby, O., Kullberg, S., Samuelsson, M.O., 1990. Chemical fluxes and mass balances in a marine fish cage farm: I. Carbon. Mar. Ecol., Prog. Ser. 61, 61 – 73. Hargrave, B.T., Duplisea, E., Pfiffer, E., Wildish, D.J., 1993. Seasonal changes in benthic fluxes of dissolved oxygen and ammonium associated with cultured Atlantic salmon. Mar. Ecol., Prog. Ser. 96, 249 – 257. Hedges, J.I., Stern, J.H., 1984. Carbon and nitrogen determinations of carbonate-containing solids. Limnol. Oceanogr. 29, 657 – 663. Holby, O., Hall, P.O.J., 1991. Chemical fluxes and mass balances in a marine fish cage farm: II. Phosphorus. Mar. Ecol., Prog. Ser. 70, 263 – 272. Holmer, M., Kristensen, E., 1992. Impact of marine fish farming on metabolism and sulphate reduction of underlying sediments. Mar. Ecol., Prog. Ser. 80, 191 – 201. Johannessen, P.J., Botnen, H.B., Tvedten, Ø.F., 1994. Macrobenthos: before, during and after a fish farm. Aquac. Fish Manage. 25, 55 – 66. Karakassis, I., Hatziyanni, E., Tsapakis, M., PlaIti, W., 1999. Benthic recovery following cessation of fish farming: a series of successes and catastrophes. Mar. Ecol., Prog. Ser. 184, 205 – 218. Kaspar, H.F., Hall, G.H., Holland, A.J., 1988. Effects of sea cage salmon farming on sediment nitrification and dissimilatory nitrate reductions. Aquaculture 70, 333 – 344. Krebs, C.J., 1972. Ecology: the Experimental Analysis of Distribution and Abundance. Harper and Row Publishers, New York. 694 pp. Lumb, C.M., 1989. Self-pollution by Scottish Salmon Farms? Mar. Pollut. Bull. 20, 375 – 379. Moore, D.C., Rodger, G.K., 1991. Recovery of a sewage sludge dumping ground: II. Macrobenthic community. Mar. Ecol., Prog. Ser. 75, 301 – 308.
330
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Nickell, L.A., Hughes, D.J.H., Atkinson, R.J.A., 1995. Megafaunal bioturbation in organically enriched Scottish sea lochs. In: Eleftheriou, A., Ansell, A., Smith, C. (Eds.), Biology and Ecology of Shallow Coastal Waters. Proceedings of the 28th European Marine Biology Symposium. 1993, Institute of Marine Biology of Crete, Iraklio, Crete, pp. 315 – 322. Nickell, L.A., Black, K.D., Hughes, D.J., Overnell, J., Brand, T., Nickell, T.D., Breuer, E., Harvey, S.M., 2003. Bioturbation, sediment fluxes and benthic community structure around a salmon cage farm in Loch Creran, Scotland. J. Exp. Mar. Biol. Ecol. 285, 221 – 233. Parkes, R.J., Buckingham, W.J., 1986. The flow of organic carbon through aerobic respiration and sulphatereduction in inshore marine sediments. In: Megusar, F., Gautar, M. (Eds.), Proceedings of the 4th International Symposium on Microbial Ecology, pp. 617 – 624. Pearson, T.H., Black, K.D., 2001. The environmental impact of marine fish cage culture. In: Black, K.D. (Ed.), Environmental Impacts of Aquaculture. Academic Press, Sheffield, pp. 1 – 31. Pearson, T.H., Rosenberg, R., 1978. Macrobenthic succession in relation to organic enrichment and pollution of the marine environment. Annu. Rev.-Oceanogr. Mar. Biol. 16, 229 – 311. Pearson, T.H., Stanley, S.O., 1979. Comparative measurements of the redox potentials of marine sediments as a rapid means of assessing the effect of organic pollution. Mar. Biol. 53, 371 – 379. Pereira, P.M.F., 1997. Macrobenthic succession and changes in sediment biogeochemistry following marine fish farm. Ph.D. Thesis. University of Stirling. 221 pp. Phillips, C.J., 1995. Effects of Atlantic Salmon farming on bacterial processes in marine sediments. PhD Thesis. University of Dundee. 167 pp. Pohle, G., Frost, B., Findlay, R., 2001. Assessment of regional benthic impact of salmon mariculture within the Letang Inlet, Bay of Fundy. ICES J. Mar. Sci. 58 (2), 417 – 426. Rigdway, I.M., Price, N.B., 1987. Geochemical associations and post-depositional mobility of heavy metals in coastal sediments: Loch Etive, Scotland. Mar. Chem. 21, 229 – 248. Ritz, D.A., Lewis, M.E., Shen, M., 1989. Response to organic enrichment of infaunal macrobenthic communities under salmonid seacages. Mar. Biol. 103, 211 – 214. Rosenberg, R., 1972. Benthic faunal recovery in a Swedish fjord following the closure of a sulphide pulp mill. Oikos 23, 92 – 108. Rosenberg, R., 1973. Succession in benthic macrofauna in a Swedish fjord subsequent to the closure of a sulphide pulp mill. Oikos 24, 244 – 258. Rosenberg, R., 1976. Benthic faunal dynamics during succession following pollution abatement in a Swedish estuary. Oikos 27, 414 – 427. SEAS, 1991. Report to Golden Sea Produce. Clyde River Purification Board Report. 93 pp. SEPA, 1992. An assessment of the impact of fish farming activities to the waters and sediments of Loch Creran, cage group A. Clyde River Purification Board Report. Weston, D.P., 1990. Quantitative examination of macrobenthic community changes along an organic enrichment gradient. Mar. Ecol., Prog. Ser. 61, 233 – 244. Westrich, J.T., Berner, R.A., 1984. The role of sedimentary organic matter in bacterial sulphate reduction: the G model tested. Limnol. Oceanogr. 29, 236 – 249. Ye, L.X., Ritz, D.A., Fenton, G.E., Lewis, M.E., 1991. Tracing the influence on sediments of organic waste from a salmonid farm using stable isotope analysis. J. Exp. Mar. Biol. Ecol. 145, 161 – 174.
Recovery of sediments after cessation of marine fish farm production Paula M.F. Pereira a,1, Kenneth D. Black a,*, Donald S. McLusky b, Thom D. Nickell a a
Dunstaffnage Marine Laboratory, Scottish Association for Marine Science, Oban, Argyll, Scotland PA37 1QA, UK b Department of Biological and Molecular Sciences, Stirling University, Stirling, Scotland FK9 4LA, UK Received 8 May 2003; received in revised form 11 December 2003; accepted 15 December 2003
Abstract Following cessation of fish production at a fish farm site in Loch Creran, Scotland, a study of the recovery of the benthic environment was undertaken. Sediment samples for macrofauna and geochemical parameters (redox potential, organic carbon, oxygen flux) were collected over a period of 15 months from three stations following a gradient of impact from the former fish farm site. The data collected were analysed by a combination of uni-and multivariate statistical methods. The macrobenthic community at the two stations furthest from the fish cage site showed signs of recovery with time in terms of indicator species, number of species and abundance, being, however, still moderately to slightly disturbed at the end of this study. At the station nearest to the former fish cage site, recovery of the macrobenthic community was also evident, but this station was still highly impacted 15 months after fish production ceased, with opportunistic species dominant. Fifteen months after fallowing, highly reduced conditions were still persistent in subsurface sediments at the stations on the periphery of the former fish cage site. Bulk sediment organic carbon, although an indicator of a spatial gradient, was not found to be a significant indicator of recovery. Combinations of different environmental parameters appear to affect different stages of benthic recovery with sediment oxygen uptake as the main observed parameter conditioning early stages of macrobenthic succession. D 2004 Elsevier B.V. All rights reserved. Keywords: Recovery; Oxygen Flux; Macrofauna; Sediment redox; Fish farm; Benthos
* Corresponding author. E-mail address: [email protected] (K.D. Black). 1 Present address: IPIMAR, Av. 5 de Outubro, 8700 Olhao, Portugal. 0044-8486/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.aquaculture.2003.12.023
316
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
1. Introduction Intensive salmon culture in sea cages often results in changes to the benthos through enrichment of sediments with organic wastes products (reviewed by Pearson and Black, 2001). Fallowing in marine fish culture involves the removal of all fish and nets from sea cage sites for different periods of time. When using this technique for disease control, production is usually interrupted for only a few months (Bron et al., 1993; FRS, 2001). Fallowing can also be used to reduce the build up of sediments beneath the cages and thus minimise any potential risk of ‘self-pollution’ (Beveridge, 1987; Lumb, 1989). This is achieved either by repositioning the cages or by alternately farming at two separate sites, usually for 12 to 24 months (one growing cycle). In Scotland, the frequency of rotation and the length of the fallowing period are at the operator’s discretion and varies considerably depending on company policy and production pressure. Regulators may also stipulate long fallow periods to allow sediments to recover where the input has been beyond the local assimilative capacity of the environment and this may allow farming to continue at an economic scale for alternate cycles. This approach is taken to ensure that sediment quality standards are met. The biogeochemical and biological processes that drive recovery are, however, poorly studied and the appropriate fallowing time is not known for any site. Based on studies of pulp mill waste, Pearson and Rosenberg (1978) suggested that changes in macrofaunal structure are similar across both spatial and temporal gradients. Despite numerous studies having being carried out to describe spatial changes of benthic community and sediment chemical conditions in response to marine fish farming (Brown et al., 1987; Ritz et al., 1989, Weston, 1990; Ye et al., 1991; Findlay et al., 1995; Karakassis et al., 1999), relatively little information is available on the recovery from those changes. Gowen et al. (1988) and Lumb (1989), studying the effects of salmon fish farming in Scottish sea loch sediments, found that no significant benthic macrofaunal recolonisation was apparent 1 year after fish production ceased. Ritz et al. (1989), however, observed a 7-week period for recovery from moderately disturbed to undisturbed sediment conditions. Slow recovery of the macrobenthic community was reported after the removal of a salmon farm in Norway, with the community still highly dominated by Capitella capitata 1 year later (Johannessen et al., 1994). In Mediterranean waters, Karakassis et al. (1999) also observed that a fish farm site, having been stocked for 6 years, had not entirely recovered 23 months after fish cages had been removed. In Canada, Pohle et al. (2001) have also found no recovery of benthic macrofauna community 1 year after fish farming ceased. Recovery studies with other sources of organic enrichment which show similar spatial gradients as fish farm wastes, such as domestic sludge waste (Eleftheriou et al., 1982; Moore and Rodger, 1991), pulp mill waste (Rosenberg, 1972, 1973, 1976; Christie and Green, 1982) and oil spills (Cabioch et al., 1982; Gray, 1982), show biological parameters (diversity, abundance and number of species) to be similar to background after 3 to 12 or more years depending on the degree of contamination and on hydrographic parameters. The aims of the present study were: (a) to describe the effect and time scale of the cessation of the input of fish farm waste on macrobenthic succession; and (b) to evaluate which biogeochemical parameters best correlated with recovery of the benthos.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
317
2. Materials and methods 2.1. Study site The sampling site was located in the vicinity of a salmon farm in Loch Creran, a fjordic sea loch on the west coast of Scotland north of Oban (Fig. 1). The loch is 12.8km long and has two main basins separated by shallow sills. The larger outer basin, in which the sampling site was located, is 9-km long and of 49-m maximum depth. The loch is sheltered from severe wave action and has a mean freshwater input of 286 106 m3 year 1 (Edwards and Sharples, 1986). The fish farm was composed of 10 cages moored together. Each cage was a square structure (25 25 m) from which a net was hung to a depth of 11 m. The site had been used for fish farming for several years with an estimated input of 169 tonnes of organic carbon into the environment (as waste food and faeces) for the 2 years previous to July 1994 when fallowing started. The site resumed production in May 1996. Near-bottom seawater temperatures at the sampling site ranged over the study period from 6 to 14 jC, and the direction of the residual current was NW ranging from 0 to 37 cm s 1 with a mean current of 3.8 cm s 1 at 2.5 m above the sediment surface (Pereira, 1997). The seabed at the three stations was silty with median grain diameter of 0.05 and a silt-clay content of 55% (Pereira, 1997). In order to select sampling stations, preliminary transects were set in July 1994 after fish farming had ceased but while the fish cages were still in place. The degree of organic enrichment was assessed by measuring redox potential profiles (Pearson and
Fig. 1. Loch Creran with inset showing the sampling site, bathymetry and the locations of stations A, B and C positions relative to the former fish cage group.
318
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Stanley, 1979). In situ observation of the sediment by divers was also taken into consideration (presence/absence of Beggiatoa spp. mats). Three stations were selected, all at 19-m water depth and following a gradient of impact from the former cage group location with station A (no Beggiatoa spp. mats; Eh (2-cm depth) = 22 mV; Eh (4-cm depth) = 30 mV) at 33 m south east from the cage, station B (Beggiatoa spp. mats present; Eh (2-cm depth) = 153 mV; Eh (4-cm depth) = 190 mV) at 3 m from the south east edge of the cage and station C (Beggiatoa spp. mats present; Eh (2-cm depth) = 191 mV; Eh (4-cm depth) = 213 mV) at the south edge of the cage, 25 m west from station B (Fig. 1). The stations were located up-current from the cage group to minimise the distance between stations, thus allowing maximum bottom time for diving and also minimising inter-station differences in depth and substratum. Due to the presence in Loch Creran of other possible inputs of high organic enrichment such as other fish farms, a fish processing plant and a kelp processing plant, it was impossible to find a reference site with the same hydrographic and sedimentary characteristics away from the influence of the fish farm. 2.2. Sampling procedures Samples were collected 1 (4/8/1994), 2 (5/9/1994), 8 (6/3/1995), 14 (21/8/1995) and 15 (11/10/1995) months after fallowing at stations A and B. Station C was sampled at all the above dates excepting 4/8/1994 when, due to adverse weather conditions, the dive was aborted. Sediment samples were collected by SCUBA divers using core tubes (57-mm i.d. 230-mm height). This sampling technique was preferred to the use of a grab due to the precise station fixing allowed by diver coring along a fixed transect, necessary in areas of steep organic enrichment gradients. At each station and each date, three replicate sediment cores were collected for analyses of benthic macrofauna. The samples were fixed in 10% formalin buffered with borax, until further analysis. The samples were washed through a 500-Am sieve and all the retained benthos hand-sorted and identified to the lowest practical taxonomic level. Species diversity was calculated by means of the (log2) Shannon –Wiener index (Krebs, 1972). The percentages of organic carbon and nitrogen were determined from duplicate cores in set intervals (0– 1; 1 –3; 3 –6; 6 –9cm). The organic carbon and nitrogen sediment samples were measured using a Leco CHN-900 Analyser following vapour-phase acidification (Hedges and Stern, 1984). Oxidation– reduction potential (Eh) was determined in duplicate cores on the overlying water 1 cm above the sediment, and from the sediment surface at 0.5-cm intervals to 7-cm depth as described by Pearson and Stanley (1979). Results are reported from the 2- and 4-cm sediment depths only, as these are considered to be the most representative of the overall levels down the sediment column (Pearson and Stanley, 1979). Sediment metabolism was measured as oxygen consumption in undisturbed sediment cores. At each sampling date, two sediment cores from each station, with their top bung removed, were stored overnight in a container with aerated water from the sampling site at in situ temperature, in order to reduce disturbance as a result of sampling and subsequent transport. The cores were then sealed with a top which incorporated an electronic magnetic stirrer to maintain a continuous water circulation at a rate just below resuspension limit, and allowing
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
319
repeated sampling of the overlying water (Parkes and Buckingham, 1986). Oxygen flux was determined from the concentration difference between initial and final samples during incubation periods of 3 h. Dissolved oxygen was determined using the micro Winkler back-titration method (Grasshoff, 1983). A Malvern Mastersizer/E and relevant software (NEC PC 486SX running Mastersizer/ E version 1.2a) were used to determine the median grain size (Md), sediment percentage silt-clay ( < 63 Am) and sediment sorting (rI) from core samples collected from each station at each date. Sea water temperature was measured with a digital diving depth meter in the water overlying the sediment. 2.3. Statistical analyses Multivariate analyses followed methods described by Clarke (1993) using the PRIMER (Plymouth Routines in Multivariate Ecological Research) software package. To satisfy statistical assumptions for significance-testing procedures, all environmental data were log10-transformed and faunal abundance data square-root transformed. The CLUSTER method was used to compute a similarity matrix based on the Bray– Curtis similarity coefficient. The similarity matrix was subjected to hierarchical, agglomerative classification employing group-average sorting to classify the stations based on faunal groupings. Non-metric Multidimensional Scaling ordination analyses (MDS) were performed on the similarity matrices obtained using the Bray– Curtis similarity coefficient. To test for statistically significant differences between the macrobenthos structures at the different stations and sampling dates, a two-way crossed ANOSIM test was used. The BIO-ENV method was used in order to demonstrate the most important environmental variables related to faunal patterns.
3. Results An extensive display of the chemical and physical results and the list of species per each core sampled with respective abundance are given in Pereira (1997). 3.1. Physical and chemical data The organic carbon content of the sediment surface layer on the edge of the cage group (station C) was about twice of those values found 55 m away (station A) (Fig 2). The sediment oxygen consumption was also elevated (Fig. 3) and the sediment redox potential was substantially depressed (Fig. 4). At the two first sampling dates white mats of the sulphur-oxidizing bacteria, Beggiatoa spp., covered the seabed at station C (Table 1). At station C the increase in organic carbon from March to August 1995 (months 8 to 14) at all depth horizons is most accentuated at the 1 –3-cm depth horizon (Fig. 2). In month 15, the organic carbon decreased in the upper sediment layers (0 –1 and 1– 3 cm), but increased in the deeper sediment layer (3– 6 cm). At station C, pockets of Beggiatoa spp. reappeared 14 and 15 months after fallowing (August and October 1995), indicative of hypoxic sediment conditions.
320
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 2. Temporal changes in organic carbon percentage in the sediment, at different depth horizons at the three stations sampled.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 3. Temporal changes in sediment oxygen flux rate (mmol d
1
m
321
2
) at the three stations sampled.
Fig. 4. Temporal changes in redox potential values (mV) at 2- and 4-cm sediment depth at the three stations sampled.
322
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Table 1 Average relative abundance (%) per station (A, B, C) and different fallowing times of the most abundant polychaete species Station
A
Months of fallowing 1 Presence of Beggiatoa spp. C.spp (ind m 2) 6 M.frag (ind m 2) 9 M.ful (ind m 2) 0 P.fallax (ind m 2) 4 39* O.hart (ind m 2) S.infl (ind m 2) 2
B 2
8
14
15
1
C 2
8
14 15 + + 9 0 0 0 26* 51* 43* 5 3 16 24* 34* 29* 14 1 5 28* 21* 0 0 0 0 16 17 0 0 1 2 11 3 13 0 0 0 15 20 41* 0 0 0 21 20 4 0 0 1 6 12 4 0 0 4 4 0
1 2 8 14 15 + + + + – 74* 56* 28* 10 – 0 3 25 14* – 17 0 0 8 – 0 0 3 1 – 1 0 0 0 – 0 2 0 0
(*) Dominant polychaete species. C. spp: Capitela spp.; M. frag: Mediomastus fragilis; M. ful: Malacocerus fuliginosus; P. fallax: Prionospio fallax; O. hart: Ophryotrocha hartmanni; S. infl: Scalibregma inflatum.
3.2. Macrofaunal data A total of 71 taxa was identified, with a minimum of 9 taxa and a maximum of 44 taxa found per station at all dates (Fig. 5). The number of species increased with distance from the fish farm site and in general with fallowing time. While station A showed a seasonal trend in macrofaunal abundance, station B and C abundance generally decreased with fallowing time (Fig. 5). At the beginning of the study, Ophryotrocha hartmanni and Mediomastus fragilis were the most abundant species present at station A (Table 1). Capitella sp. was the third most abundant polychaete accounting for 6– 9% of the total abundance of that station. Malacocerus fuliginosus was never present in station A samples. M. fragilis and Prionospio fallax accounted for 42% of the total abundance 15 months after fallowing at station A. Capitella sp., M. fuliginosus and O. hartmanni were the dominant species (64% and 88% at Station B and 91% at Station C) at Stations B and C at the two first sampling dates. Fifteen months after fallowing, M. fragilis and P. fallax were the most abundant polychaete species at station B. M. fragilis was found in smaller numbers at station C than at the other two stations but showed a steady increase over time and was the annelid with the highest density 15 months after fallowing, closely followed by Capitella sp. and M. fuliginosus (Table 1). The Shannon – Wiener diversity index at station A fluctuated between 3.1 and 3.7 (Fig. 6). The species diversity in station B increased until it reached 3.6, 14 months after fallowing started, decreasing to 3.0 at 15 months. The species diversity at station C, although increasing with fallowing time, was consistently lower than 3.0 throughout the survey. 3.3. Multivariate statistics The multidimensional scaling configuration (MDS) for the abundance data (Fig. 7) shows that the three stations follow a similar nonlinear temporal trajectory. The rate of change was different between stations, with station B macrofaunal structure changing
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 5. Temporal changes in total number of taxa (in 0.0075 m2) and mean abundance (ind m stations sampled.
323
2
) at the three
Fig. 6. Temporal changes in species diversity (log2 Shannon – Wiener index) at the three stations sampled.
324
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Fig. 7. MDS configuration of square-root transformed species-abundance mean data at each station (A, B, C) and at each sampling data (1, 2, 8, 14 and 15 months of fallowing).
(recovering) at a higher rate than station C. Although there was a similar tendency for all stations to move towards recovery, all stations were found to be significantly different from each other at the 5% level (Table 2). Again, all pairs of dates were found to be significantly different with dissimilarity generally decreasing in time (Table 2). The BIO-ENV multivariate showed the combination of particle diameter, oxygen flux and organic carbon variables as the match that best grouped the stations in a manner consistent with the faunal pattern. The match, however, was not very successful (0.57) Table 2 Two-way crossed ANOSIM test for differences between (a) stations and (b) sampling dates, using square-root transformed species abundance data, and the Bray – Curtis similarity measure (three replicates at each station) (a) Tests for differences between stations groups Sample statistic (Global R): 0.770 Significance level of sample statistic: 0.001% Groups used
Statistical value R
Significance level (%)
(A, B) (A, C) (B, C)
0.815 0.991 0.657
0.001 0.001 0.001
(b) Tests for differences between time groups (as months of following) Sample statistic (Global R): 0.806 Significance level of sample statistic: 0.001% Groups used
Statistical value R
Significance level (%)
2, 8 2, 14 2, 15 8, 14 8, 15 14, 15
0.938 1.000 1.000 0.617 0.728 0.519
0.001 0.001 0.001 0.001 0.001 0.001
The samples from stations A and B in August 1994 were not analysed as there were no comparative data from station C.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
325
Table 3 Combinations of the 7 environmental variables (organic carbon, oxygen flux, redox potential, particle diameter, silt-clay percentage, sediment sorting, sea water temperature) yielding the best match of biotic and abiotic similarity matrices, as measured by weighted Spearman rank correlation qN; correlation coefficient in brackets BIO-ENV; weighted Spearman rank correlation
Best variable combination
Station A
Station B
Station C
O2, OC (0.66)
O2 (0.81)
O2 (0.94)
O2—oxygen flux; OC—organic carbon.
indicating that the environmental variables measured only partially explain the distribution of the faunal data. Since the stations sampled reflect different influences from the fish farm waste (due to the distance from the source), it was thought that the weight of the different environmental variables ‘explaining’ the community structure at the different stations might also be distinct. Table 3 shows the single and combination of environmental variables that best group the temporal variation of individual stations in a manner consistent with the faunal pattern. The correlation coefficient decreases with distance from the fish farm site. No physical variable (sea water temperature, sediment particle size, sorting coefficient, sediment clay fraction) was found to be included in the best match. Oxygen consumption was found to be the environmental parameter that better correlated with the faunal variation in time at stations B and C. At station A the combination of oxygen and concentration of organic carbon in the sediment showed the best correlation with the faunal pattern.
4. Discussion The degree and types of disturbance observed at the fish farm site in Loch Creran were broadly consistent with other studies on effects of fish farm waste on sediment chemical and biological properties (Brown et al., 1987; Kaspar et al., 1988; Ritz et al., 1989; Weston, 1990, Holmer and Kristensen, 1992; Karakassis et al., 1999). Although seasonal variations in ‘natural’ organic material input, sedimentation, temperature and larval availability will have an additional effect on the environmental and biological parameters measured, the general worsening of sediment and biological conditions with proximity to the fish farm site, and the gradual improvement with fallowing time, suggests that the effects found were mainly related to the cessation of input of fish farm waste. The consumption of oxygen by chemical or biological processes could not be distinguished using the data collected. The high variability between replicate measurements of oxygen uptake could reflect spatial patchiness in sedimentation of faecal waste and food pellets and also the patchy effects of epi-faunal activities. The initial oxygen uptake values of 80– 100 mmol m 2 d 1 were typical of enriched sediments (Holby and Hall, 1991; Hargrave et al., 1993; Phillips, 1995). Much higher oxygen flux rates (435 mmol m 2 day 1;Nickell et al., 2003) have subsequently been measured under a different farm in Loch Creran, but this was during a period of high production rather than ca. 1 month after production ceased as in the first samples in the present study. In March 1995, 8 months after fallowing, oxygen uptake rates (30 –50 mmol m 2 d 1) were still greater than estimates for sea-lochs not subjected to fish farming (5 – 25 mmol
326
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
m 2 d 1) (Davies, 1975; Parkes and Buckingham, 1986; Hall et al., 1990; Nickell et al., 2003). Organic material available to the benthos can be divided into two decomposable fractions of considerably different reactivity plus a non-metabolizable fraction (Westrich and Berner, 1984; Grant and Hargrave, 1987). The initial high oxygen consumption is likely to be due to the residual high levels of highly degradable organic carbon originating from the fish farm. The percentage of organic carbon in the surface sediment (0 – 1 cm) observed throughout the survey at stations B and C (2– 4%) was within the range of background levels found in other sea-lochs (Rigdway and Price, 1987; Nickell et al., 1995) and similar to values observed for Loch Creran (2 –5%) before the introduction of fish farms (Ansell, 1974) and to previous years’ background values when the fish farm was active (SEAS, 1991; SEPA, 1992). However, the percentage of organic carbon in the sediment (0– 3 cm) decreased with distance from the fish farm site, being two to three times higher at the edges of the cages (stations B and C) than 30 m away (station A); similar results being found at other fish farm sites (Brown et al., 1987; Hall et al., 1990; Weston, 1990; Ye et al., 1991; Holmer and Kristensen, 1992; Karakassis et al., 1999). The decrease of organic carbon in the surface sediment at stations B and C after fish farm production ceased, approaching the values at station A, is also an indication that the fish farm waste was the main origin of the organic carbon present in the sediment at the beginning of the study. The presence of elevated organic carbon, relative to station A, at stations B and C to a depth of 6 cm, 15 months into fallowing, indicates the persistence of contamination by fish farm waste in these subsurface sediments, consistent with low redox potential values found at the same depths. Moore and Rodger (1991) noted that although the first layers of sediment showed signs of recovery, deeper sediments remained contaminated for longer. This could affect the degree and time of response to further disturbances, with natural disturbances having a greater effect than in non depth-contaminated sediments. Seasonal variations in sediment redox potential at stations B and C followed seasonal temperature changes, indicating an increased sediment metabolism during the summer months of maximum temperature but the variables were not significantly correlated as has been found in previous fish farm studies (Holmer and Kristensen, 1992; Hargrave et al., 1993). Most biological univariate indices used illustrated gradual faunal changes over time, from highly disturbed (stations B and C) to moderately disturbed conditions. The seasonal influence, however, appears to have a strong effect over recovery with all these parameters regressing over the summer months, 14 months into fallowing. This effect is represented by the nonlinear trajectory of the MDS distribution of the individual stations with time (Fig. 7) where initial recovery towards the upper left is succeeded by a regression downwards for all stations (and subsequently to the right for stations A and C at the final sampling). Karakassis et al. (1999) related the regression of recovery observed in their study to the increase of degrading organic material buried after the peak of benthic algal production due to pulses of phosphorous released after intense fluctuations between reduced and oxidised conditions in the sediment after cessation of fish farming. Scottish sea lochs are typically turbid and benthic algal production is insignificant at the depths of the stations studied here; therefore, the
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
327
similar trends observed in the present study must be explained by fresh inputs of natural organic material (river runoff and phytoplankton production) and increased summer water temperature rather than stimulated in situ production. Varying degrees of recovery were observed at the stations studied, indicating that the speed of recovery decreases with increasing fallowing time. Although stations B and C were both highly disturbed at the beginning of the study as indicated by the high dominance by opportunistic species, the times between the successive stages of recovery were different for both stations, with station B recovering more rapidly than station C as indicated by the species succession and univariate and multivariate results. Fifteen months after the fish farm loading ceased, the highly disturbed benthic structure at station C, although improving, was still dominated by opportunistic species characteristic of highly disturbed sediments, such as Capitella sp, M. fragilis and M. fuliginosus. After 15 months of fallowing, the benthic structure at station B was characteristic of a moderately disturbed community. Station A, with a benthic structure characteristic of moderately disturbed sediments at the beginning of the study, still showed indications of being disturbed 15 months after fallowing. Lumb (1989) and Gowen et al. (1988), studying the effects of fish farms waste on Scottish sea loch sediments, found results similar to station C, with the sediment still organically enriched and no significant macrofaunal recolonisation appearing to have occurred 1 year after fish production had ceased. Slow recovery of the macrobenthic community was also reported after the removal of a salmon farm in Norway, with the community still highly dominated by C. capitata 1 year later (Johannessen et al., 1994). In Canada, Pohle et al. (2001) found no recovery of the benthic macrofauna community 1 year after fish farming ceased. Hargrave et al. (1993) compared methods for detecting organic impacts near salmon farms on the east coast of Canada and showed that total sulphide, benthic O2 and CO2 exchange, and redox potential were the most sensitive measures of 20 variables tested. When excluding site specific properties by analysing temporal variations at the individual stations, the correlation coefficient between oxygen uptake and macrobenthic fauna increased with proximity to the waste source, with this parameter appearing to be the main factor associated with macrobenthic changes at the two stations near to the fish farm cages (stations B and C). The decrease in correlation coefficient with distance from the fish farm site, and the fact that different groups of environmental parameters were found to correlate with faunal distribution at the different stations, indicates that the significance of individual environmental parameters varies with the degree of disturbance. This is likely to be a complex feedback between the microbial degradation of organic matter and sediment colonisation by tolerant, bioturbating macroorganisms that act to increase the flux of electron acceptors in the sediment while themselves contributing to the overall sediment oxygen demand. Although the amount of organic matter is believed to be the main factor influencing macrofaunal community structure after deposition of organic enriched waste (Pearson and Rosenberg, 1978; Moore and Rodger, 1991), the percentage of organic carbon in the sediment was not found in this study to be a good indicator of temporal changes in the macrobenthic community. This is in accordance with other author’s results (Dauer and Conner, 1980; Ye et al., 1991; SEPA, 1992) and it may be a result, in part, of the fact that the correlation between organic carbon and community changes depends on the labile fraction of the organic
328
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
matter (Grant and Hargrave, 1987) and not on the bulk sediment organic carbon measured in the present and previous studies, and also, in part, on the burial processes. Although at station A most sediment chemical variables such as redox potential and organic carbon showed levels characteristic of unenriched sediments, the macrobenthos structure indicated a moderate disturbance. In a recovery study from sludge waste, Eleftheriou et al. (1982) also found that 8 months after sludge addition ceased, the sediment (organic carbon and redox potential) had returned to control levels but the fauna (diversity, abundance and species numbers) had not. The same effect was observed on a spatial scale in active fish farms (Brown et al., 1987; Weston, 1990; SEPA, 1992) with effects of fish farm waste on the benthic community still apparent even when sediment chemistry (redox potential, organic carbon, water-soluble sulphide) was similar to background levels. On a spatial scale, this effect indicates that the fauna are sensitive to enrichment at levels not detectable with the gross chemical measures used. On a temporal scale, however, we also need to take into consideration that the recovery of the fauna should be slower than the recovery of the chemical conditions since the recruitment of less pollution-tolerant species can only take place after reestablishment of the environmental conditions. In Scotland, fallowing of fish farm sites is recommended by the regulator for the purpose of disease control (SOAEFD, 1996), but using this technique to avoid the build up of organic wastes in sediments is more controversial. Improvement of management techniques and diets minimises waste and increases food efficiency, so that less carbon, nitrogen and other nutrients reach the sediment beneath fish cages. The present study, however, suggests that a 2-year rotation period is unlikely to lead to the complete recovery of the most affected sediment and its associated benthic community to predisturbance conditions. Although some recovery of the benthic community takes place during fallowing, contamination in subsurface sediments persists, suggesting a rapid return to pre-fallowing conditions as soon as production resumes. For a company to be able to fallow a site while maintaining fish production, new sites would need to be granted thus increasing the total seabed surface affected by fish farming.
Acknowledgements This research was supported by the Portuguese Government, Junta Nacional de Investigacß a˜o Cientı´fica e Tecnolo´gica, Sub-Programa Cieˆncia e Tecnologia do 2j Quadro Comunita´rio de Apoio. We are grateful for the assistance provided by the staff of Scottish Seafarms and in particular by Sally Davies, by the crews of the SAMS research vessels and by the SAMS diving team especially Paul Provost.
References Ansell, A.D., 1974. Sedimentation of organic detritus in Lochs Etive and Creran, Argyll, Scotland. Mar. Biol. 27, 263 – 273.
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
329
Beveridge, M.C.M., 1987. Cage Aquaculture. Fishing News Books, Farnham, Surrey. Bron, J.E., Sommerville, C., Wootten, R., Rae, G.H., 1993. Following of marine Atlantic salmon, Salmo salar L., farms as a method for the control of sea lice, Lepeophtheirus salmonis (Kroyer, 1837). J. Fish Dis. 16, 487 – 493. Brown, J.R., Gowen, R.J., McLusky, D.S., 1987. The effect of salmon farming on the benthos of a Scottish sea loch. J. Exp. Mar. Biol. Ecol. 109, 39 – 51. Cabioch, L., Dauvin, J.C., Retiere, C., Rivain, V., Archambault, D., 1982. Evolution de peuplements benthiques des fonds sedimentaires de la region de Roscoff, perturbes par les hydrocarbures de L’Amoco Cadiz. Neth. J. Sea Res. 16, 491 – 501. Christie, H., Green, N.W., 1982. Changes in the sublittoral hard bottom benthos after a large reduction in pulp mill waste to Iddefjord, Norway, Sweden. Neth. J. Sea Res. 16, 474 – 482. Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18, 117 – 143. Dauer, D.M., Conner, W.G., 1980. Effects of moderate sewage input on benthic polychaete populations. Estuar. Coast. Mar. Sci. 10, 335 – 346. Davies, I.M., 1975. Energy flow through the benthos in a Scottish sea loch. Mar. Biol. 31, 353 – 362. Edwards, A., Sharples, F., 1986. Scottish Sea Lochs: A Catalogue. Scottish Marine Biological Association, Oban. 210 pp. Eleftheriou, A., Moore, D.C., Basford, D.J., Robertson, M.R., 1982. Underwater experiments on the effects of sewage sludge on a marine ecosystem. Neth. J. Sea Res. 16, 465 – 473. Findlay, R.H., Watling, L., Mayer, L.M., 1995. Environmental impact of salmon net-pen culture on Maine marine benthic communities: a case study. Estuaries 18, 145 – 179. FRS, 2001. Scottish fish farms, annual production survey. The Scottish Executive Environment and Rural Affairs Department, Aberdeen Marine Laboratory. 47 pp. Gowen, R.J., Brown, J.R., Bradbury, N.B., McLusky, D.S., 1988. Investigations into benthic enrichment, hypernutrification and eutrophication associated with mariculture in Scottish coastal waters (1984 – 1988). Department of Biological Sciences, University of Stirling. 289 pp. Grant, J., Hargrave, B.T., 1987. Benthic metabolism and the quality of sediment organic carbon. Biol. Oceanogr. 4, 243 – 264. Grasshoff, K., 1983. Methods of Seawater Analysis, 2nd ed. Verlag Chemie, Weinheim. 419 pp. Gray, J.S., 1982. Effect of pollutants on marine ecosystems. Neth. J. Sea Res. 16, 424 – 443. Hall, P.O., Anderson, L.G., Holby, O., Kullberg, S., Samuelsson, M.O., 1990. Chemical fluxes and mass balances in a marine fish cage farm: I. Carbon. Mar. Ecol., Prog. Ser. 61, 61 – 73. Hargrave, B.T., Duplisea, E., Pfiffer, E., Wildish, D.J., 1993. Seasonal changes in benthic fluxes of dissolved oxygen and ammonium associated with cultured Atlantic salmon. Mar. Ecol., Prog. Ser. 96, 249 – 257. Hedges, J.I., Stern, J.H., 1984. Carbon and nitrogen determinations of carbonate-containing solids. Limnol. Oceanogr. 29, 657 – 663. Holby, O., Hall, P.O.J., 1991. Chemical fluxes and mass balances in a marine fish cage farm: II. Phosphorus. Mar. Ecol., Prog. Ser. 70, 263 – 272. Holmer, M., Kristensen, E., 1992. Impact of marine fish farming on metabolism and sulphate reduction of underlying sediments. Mar. Ecol., Prog. Ser. 80, 191 – 201. Johannessen, P.J., Botnen, H.B., Tvedten, Ø.F., 1994. Macrobenthos: before, during and after a fish farm. Aquac. Fish Manage. 25, 55 – 66. Karakassis, I., Hatziyanni, E., Tsapakis, M., PlaIti, W., 1999. Benthic recovery following cessation of fish farming: a series of successes and catastrophes. Mar. Ecol., Prog. Ser. 184, 205 – 218. Kaspar, H.F., Hall, G.H., Holland, A.J., 1988. Effects of sea cage salmon farming on sediment nitrification and dissimilatory nitrate reductions. Aquaculture 70, 333 – 344. Krebs, C.J., 1972. Ecology: the Experimental Analysis of Distribution and Abundance. Harper and Row Publishers, New York. 694 pp. Lumb, C.M., 1989. Self-pollution by Scottish Salmon Farms? Mar. Pollut. Bull. 20, 375 – 379. Moore, D.C., Rodger, G.K., 1991. Recovery of a sewage sludge dumping ground: II. Macrobenthic community. Mar. Ecol., Prog. Ser. 75, 301 – 308.
330
P.M.F. Pereira et al. / Aquaculture 235 (2004) 315–330
Nickell, L.A., Hughes, D.J.H., Atkinson, R.J.A., 1995. Megafaunal bioturbation in organically enriched Scottish sea lochs. In: Eleftheriou, A., Ansell, A., Smith, C. (Eds.), Biology and Ecology of Shallow Coastal Waters. Proceedings of the 28th European Marine Biology Symposium. 1993, Institute of Marine Biology of Crete, Iraklio, Crete, pp. 315 – 322. Nickell, L.A., Black, K.D., Hughes, D.J., Overnell, J., Brand, T., Nickell, T.D., Breuer, E., Harvey, S.M., 2003. Bioturbation, sediment fluxes and benthic community structure around a salmon cage farm in Loch Creran, Scotland. J. Exp. Mar. Biol. Ecol. 285, 221 – 233. Parkes, R.J., Buckingham, W.J., 1986. The flow of organic carbon through aerobic respiration and sulphatereduction in inshore marine sediments. In: Megusar, F., Gautar, M. (Eds.), Proceedings of the 4th International Symposium on Microbial Ecology, pp. 617 – 624. Pearson, T.H., Black, K.D., 2001. The environmental impact of marine fish cage culture. In: Black, K.D. (Ed.), Environmental Impacts of Aquaculture. Academic Press, Sheffield, pp. 1 – 31. Pearson, T.H., Rosenberg, R., 1978. Macrobenthic succession in relation to organic enrichment and pollution of the marine environment. Annu. Rev.-Oceanogr. Mar. Biol. 16, 229 – 311. Pearson, T.H., Stanley, S.O., 1979. Comparative measurements of the redox potentials of marine sediments as a rapid means of assessing the effect of organic pollution. Mar. Biol. 53, 371 – 379. Pereira, P.M.F., 1997. Macrobenthic succession and changes in sediment biogeochemistry following marine fish farm. Ph.D. Thesis. University of Stirling. 221 pp. Phillips, C.J., 1995. Effects of Atlantic Salmon farming on bacterial processes in marine sediments. PhD Thesis. University of Dundee. 167 pp. Pohle, G., Frost, B., Findlay, R., 2001. Assessment of regional benthic impact of salmon mariculture within the Letang Inlet, Bay of Fundy. ICES J. Mar. Sci. 58 (2), 417 – 426. Rigdway, I.M., Price, N.B., 1987. Geochemical associations and post-depositional mobility of heavy metals in coastal sediments: Loch Etive, Scotland. Mar. Chem. 21, 229 – 248. Ritz, D.A., Lewis, M.E., Shen, M., 1989. Response to organic enrichment of infaunal macrobenthic communities under salmonid seacages. Mar. Biol. 103, 211 – 214. Rosenberg, R., 1972. Benthic faunal recovery in a Swedish fjord following the closure of a sulphide pulp mill. Oikos 23, 92 – 108. Rosenberg, R., 1973. Succession in benthic macrofauna in a Swedish fjord subsequent to the closure of a sulphide pulp mill. Oikos 24, 244 – 258. Rosenberg, R., 1976. Benthic faunal dynamics during succession following pollution abatement in a Swedish estuary. Oikos 27, 414 – 427. SEAS, 1991. Report to Golden Sea Produce. Clyde River Purification Board Report. 93 pp. SEPA, 1992. An assessment of the impact of fish farming activities to the waters and sediments of Loch Creran, cage group A. Clyde River Purification Board Report. Weston, D.P., 1990. Quantitative examination of macrobenthic community changes along an organic enrichment gradient. Mar. Ecol., Prog. Ser. 61, 233 – 244. Westrich, J.T., Berner, R.A., 1984. The role of sedimentary organic matter in bacterial sulphate reduction: the G model tested. Limnol. Oceanogr. 29, 236 – 249. Ye, L.X., Ritz, D.A., Fenton, G.E., Lewis, M.E., 1991. Tracing the influence on sediments of organic waste from a salmonid farm using stable isotope analysis. J. Exp. Mar. Biol. Ecol. 145, 161 – 174.