Environ. Sci. Technol. 2010, 44, 2079–2084
Factors Affecting Benthic Impacts at Scottish Fish Farms D A N I E L J . M A Y O R , * ,† A L A I N F . Z U U R , ‡ MARTIN SOLAN,† GRAEME I. PATON,§ AND KEN KILLHAM§ Institute of Biological and Environmental Sciences, Oceanlab, University of Aberdeen, Main Street, Newburgh, Aberdeenshire, Scotland AB41 6AA, U.K., and Highland Statistics Ltd., 6 Laverock Road, Newburgh, Aberdeenshire AB41 6FN, U.K., Institute of Biological and Environmental Sciences, Cruickshank Building, University of Aberdeen, St. Machar Drive, Aberdeen AB24 3UU, U.K.
Received October 8, 2009. Revised manuscript received January 29, 2010. Accepted February 3, 2010.
The factors affecting patterns of benthic [seabed] biology and chemistry around 50 Scottish fish farms were investigated using linear mixed-effects models that account for inherent correlations between observations from the same farm. The abundance of benthic macrofauna and sediment concentrations of organic carbon were both influenced by a significant, albeit weak, interaction between farm size, defined as the maximum weight of fish permitted on site at any one time, and current speed. Above a farm size threshold of between 800 and 1000 t, the magnitude of effects at farms located in areas of elevated current speeds were greater than at equivalent farms located in more quiescent waters. Sediment concentrations of total organic matter were influenced by an interaction between distance and depth, indicating that wind-driven resuspension events may help reduce the accumulation of organic waste at farms located in shallow waters. The analyses presented here demonstrate that the production and subsequent fate of organic waste at fish farms is more complex than is often assumed; in isolation, current speed, water depth, and farm size are not necessarily good predictors of benthic impact.
Introduction Intensive fish farming activities generate a localized gradient of organic enrichment in the underlying and adjacent sediments as a result of uneaten food and faeces (1-5). The immediate ‘environmental footprint’ of a fish farm is discernible using an array of chemical and biological parameters (6) and can also be detected acoustically (7). Chemical indices, such as total sediment concentrations of organic carbon (Corg) or organic matter (Torg), are frequently determined as surrogates for levels of organic enrichment, although quantitative analysis of benthic macrofauna is widely regarded as the most sensitive, albeit expensive, means to examine the environmental effects of farming activities (1, 2, 4, 8, 9). Gradients of organic enrichment strongly * Corresponding author phone: +44 (0)1224 274401; fax: +44 (0)1224 274402; e-mail:
[email protected]. † Institute of Biological and Environmental Sciences, Oceanlab, University of Aberdeen. ‡ Highland Statistics Ltd. § Institute of Biological and Environmental Sciences, University of Aberdeen. 10.1021/es903073h
2010 American Chemical Society
Published on Web 02/23/2010
influence the abundance and diversity of infaunal communities (10), although the spatial and temporal scales over which these changes occur can differ widely depending on site-specific farming practices and local environmental conditions (2, 11-13). Nevertheless, studies examining the spatial extent of fish farming impacts generally report that their effects on the benthic environment rapidly dissipate with increasing distance from the cage edge and are typically not discernible at distances >200 m (2, 9, 14-16). Little is known about how the size of a fish farm relates to the magnitude and spatial extent of its benthic impacts. Salmon has been farmed in Scotland for decades, and this industry has expanded from an artisanal-level to one that employs thousands of people and produces >120,000 tonnes of fish annually (17, 18). The regulatory framework that underpins this industry is internationally regarded as a benchmark standard. The permissible size of an individual salmon farm in Scotland, defined as the maximum weight of fish permitted on site at any one time (maximum consented biomass, MCB; ref 19), is independently determined by the Scottish Environment Protection Agency (SEPA). It is accepted that fish farming activities will cause some impact on the seabed within a fixed ‘allowable zone of effects’ (AZE). The consenting process was recently facilitated by the introduction of the predictive mathematical model autoDEPOMOD (20). This determines a suitable MCB for any particular location given a range of environmental parameters and information on farming practices. It also determines a sitespecific size and shape for the AZE, enabling a farm’s environmental impact to be monitored. Prior to the implementation of autoDEPOMOD in late 2005, site-specific factors such as hydrography, bathymetry, presence of other fish farms, and local conservational designations were used to determine the MCB that a site can hold in an environmentally sustainably manner, i.e. without breaching SEPA quality standards (19). The AZE of a farm consented before 2005 was permitted to extend 25 m from the cage in all directions (19), and the MCB for a given location was limited to ensure that minimum sediment quality criteria within the AZE were maintained (see Table A1.7 in ref 19). The expansion of the Scottish fish farming industry has been accompanied by a growth in environmental legislation and the necessity for site operators to collect and report annual seabed monitoring information to SEPA (21), providing a unique opportunity to examine the relationship between farm size and benthic impact. Here we use data from SEPA’s national archive to investigate the environmental factors, and their interactions, that exert a discernible effect on patterns of sediment biology and chemistry around 50 Scottish fish farms.
Materials and Methods Data Extraction. All benthic monitoring data submitted by site operators to SEPA in digital format during 2005, and the corresponding hydrographic data, were extracted from the national archive (n ) 59 farms). All farms were consented prior to the introduction of autoDEPOMOD and could therefore have an AZE extending to 25 m in all directions from the cage edge (19). Fifty of these farms reported information on benthic biology (abundance and ShannonWeiner diversity, H′), typically at 0, 25, 50, and 150 m from the cage edge along one or two sampling transects (in line with, and/or perpendicular to, the predominant current direction) and ranged in size between 150 and 1999 tonnes. A subset of 21 reports also presented sediment concentrations of organic carbon (Corg), with 11 of these also including VOL. 44, NO. 6, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2079
information on sediment nitrogen (N) concentrations. An independent subset of 15 reports presented sediment concentrations of total organics (Torg), determined as the mass of material lost on ignition. The following explanatory variables were extracted from each report: MCB; tonnage of fish on site at the time of survey (TFS); sample depth in meters (depth); transect identity (transect 1 or 2); sample distance from cage edge in meters (distance; 0, 25, 50, 150); and mean current speeds (m s-1), averaged over 15 days at the following depths; 2-3 m beneath the surface (top CS), 2-3 m above the seabed (bottom CS), the middle of the water column (middle CS). An overall ‘water column’ current speed (WCCS) was calculated from these data. Each farm was ascribed a unique identity (farm ID) to allow any sitespecific effects to be identified. Six of the 265 data points were excluded from the analysis because they either formed unique observations or because they did not have accompanying hydrographic data. The median age of the farms analyzed herein was 2.7 years, and all but two were 501 - 1000 t (n ) 29), and >1000 t (n ) 12), respectively (sensu ref 19). All statistical analyses were conducted using the ‘nlme’ package (29) within the ‘R’ statistical and programming environment (30).
Results Abundance of Benthic Invertebrates (Text S2). The OM describing abundance of benthic invertebrates was a LME model that allowed for unequal variance with distance (Table 1; Figure S2). Abundance was primarily affected by the distance from the cage edge (L-ratio ) 30.19, df ) 3, p < 0.001; Figure S3), although this was dependent upon a significant, albeit weak, interaction between MCB and WCCS (L-ratio ) 4.24, df ) 1, p ) 0.040; Figure 1; Figure S4). The abundance of benthic invertebrates was predicted to decrease with increasing WCCS at farms where MCB < ∼800 tonnes, whereas the opposite effect was observed at larger farms. 2080
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 6, 2010
TABLE 1. Summary of Fixed-, Random- (r), and Variance Covariate- (β) Terms Remaining in the Optimal Models model
response
model term
df
L-ratio
p
M1
abundance
M2
H’
M3
Corg
M4
N
M5
Torg
distance MCB:WCCS farm IDR distanceβ distance farm IDR distanceβ MCB: WCCS farm IDR distance WCCS farm IDR distance:depth farm IDR
3 1 1 3 3 1 3 1 1 3 1 1 3 1
30.19 4.24 102.53 31.48 118.33 25.66 8.60 4.12 52.12 9.83 8.84 13.42 8.45 45.46
