Poems

The environmental impact of Mediterranean cage fish farms at semi-exposed locations: does it need a re-assessment?

Description
The environmental impact of Mediterranean cage fish farms at semi-exposed locations: does it need a re-assessment?
Categories
Published
of 15
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Transcript
  ORIGINAL ARTICLE Manuel Maldonado  Æ  M. Carmen. CarmonaYolanda Echeverrı ´a  Æ  Ana Riesgo  The environmental impact of Mediterranean cage fish farmsat semi-exposed locations: does it need a re-assessment? Received: 14 July 2004/ Revised: 18 November 2004/ Accepted: 18 November 2004/Published online: 25 January 2005   Springer-Verlag and AWI 2005 Abstract  During spring and summer 2003, we measureda variety of chemical and biological parameters in fivemedium-sized, Mediterranean cage farms that exploitsemi-offshore conditions, and controlled the supply of fodder. The objective was to assess whether modern cagefarms proliferating at semi-offshore sites exert environ-mental impact levels equivalent to the levels describedfrom more traditional cage farms located in shallow,sheltered sites. In the water column, we examined theconcentration of dissolved inorganic nutrients and het-erotrophic bacteria in both surface and near-bottomwater. At the bottom, we examined the concentrationsof benthic chlorophyll  a , phaeophytin and organicmatter in sediments, the granulometric structure of thesediment, and the taxonomic (at the family level)abundance of benthic macroinvertebrates. For mostparameters, we found no substantial differences betweenfarm and control sites. Rather, most variation was ex-plained as a function of depth (surface versus bottomwater) or season (spring versus summer conditions).Deviations of farm values from control values, whenthey occurred, were small and did not indicate any sig-nificant impact on either bacterioplankton or benthicchlorophyll. Only one of the five farms studied exerted adetectable impact on the benthic macroinvertebratecommunity immediately under the cages. These resultssuggest that medium-sized fish farms located on semi-exposed western Mediterranean coasts have fewer envi-ronmental impacts than traditional fish farms located inshallow, sheltered sites. Impact characterization in thesenew farms may require refinement of the standardapproach to deal with rapid dispersal of effluents andsub-lethal levels of environmental disturbance. Keywords  Aquaculture pollution  Æ  Bacterioplankton  Æ Cage fish farms  Æ  Dissolved nutrients  Æ Macroinvertebrates Introduction Since floating cages for intensive salmon aquaculturestarted proliferating in North Atlantic waters in theearly 1980s, marine ecologists have been concerned withthe environmental impact of these installations. Studieson pioneering salmon farms located in fjords, lochs, andsemi-enclosed bays on the Atlantic coasts of Europe andNorth America, and in the Pacific, soon revealed thatcage farms generate both particulate organic wastes andsoluble inorganic wastes (e.g., Brown et al. 1987; Ritzet al. 1990; Black 1991; Persson 1991; Gowen and Ezzi 1992; Hargrave et al. 1993; Gowen 1994; Ackefors et al. 1994; Findlay and Watling 1994; GESAMP 1996). Particulate organic materials (mostly fecal material anduneaten fodder) settle to the sea floor, forming darksediments characterized by high levels of organic matter,nitrogen, and phosphorous, and reduced sulfur com-pounds. This sediment is a suitable substratum forbacterial growth, which in extreme cases induces severeoxygen depletion in sediment and bottom waters (e.g.,Tsutsumi and Kikuchi 1983; Brown et al. 1987). In addition, sustained input of phosphorous and severalnitrogen compounds from farming installations mayalter natural concentration ratios of basic nutrients inthe water column at the local scale (e.g., Gowen and Ezzi1992), sometimes favoring local algal blooms andeutrophication processes (e.g., Persson 1991).It has recently been suggested that Mediterranean seabass and sea bream farms have a somewhat lesserenvironmental impact than Atlantic salmon farms (Ka-rakassis 2001). Such a suggestion is based on data ob-tained from a limited number of farms, which in mostcases are located in the eastern-Mediterranean basin(e.g., Karakassis et al. 1998, 1999, 2000, 2001; Pitta et al. Communicated by H.-D. FrankeM. Maldonado ( & )  Æ  M. C. Carmona  Æ  Y. Echeverrı ´a  Æ  A. RiesgoCentro de Estudios Avanzados de Blanes (CSIC),Acceso Cala St. Francesc 14, Blanes, 17300 Girona, SpainE-mail: maldonado@ceab.csic.esTel.: +34-972-336101Fax: +34-973-337806Helgol Mar Res (2005) 59: 121–135DOI 10.1007/s10152-004-0211-5  1999) because farming of finfish species has traditionallybeen dominated by Egypt, Greece, and Turkey. Similarto Atlantic and Pacific salmon farms, most of theseeastern-Mediterranean sea bass and sea bream farms arelocated at shallow sheltered sites and semi-enclosedbays, a location choice favored by the complex coastalstructure of Greece, Croatia, and Turkey (Basurco 2000;Bravo and Montan ˜es 2001). The situation is quite dif-ferent for most fish farms that have been installed insome western Mediterranean countries in the last decade.Progressive advances in cage building now facilitatemooring of cage farms on relatively deep bottoms andexposed sites. Indeed, about 50% of the 45 productionunitsoperatinginSpainin1998hadcagesinsemi-offshoreand offshore conditions (Basurco and Larrazabal 2000).Fish farms in semi-exposed conditions are prolifer-ating in the western Mediterranean on the assumptionthat enhanced water renewal in cages provides improvedculturing conditions. Increased fish production is ex-pected to compensate for greater monetary investmentsand higher damage risks. From an ecological point of view, it is assumed that enhanced water renewal in semi-exposed conditions may result in less environmentalimpact than is found in farms in semi-enclosed bays. If so, the environmental problems described in the litera-ture for more traditional cage farms may not reflect thetraits of the impact exerted by Mediterranean fish farmsat semi-exposed sites appropriately. To contribute to theclarification of this issue, we have selected five semi-ex-posed cage farms and have assessed their environmentalimpact by examining diverse standard water-column andbenthic parameters. Methods Farm featuresWe studied five semi-exposed cage farms scattered over900 km along the Mediterranean coast of Spain (Fig. 1),spanning a wide range of hydrographic and environ-mental conditions. All farms grow sea bream ( Sparusaurata ) and/or sea bass ( Dicentrarchus labrax ), butaverage fish biomass per farm during 2002 ranged fromto 40 to 549 tons, depending on farm (Table 1). There-fore, the study considers a variety of conditions, notonly environmental but also farming. Most farms wereless than 6 years old, so that they can be consideredmodern installations. They all use either 19 or 25 mdiameter floating cages moored on soft bottoms between21 and 37 m deep (Table 1). Fish feeding is monitoredby underwater video cameras, allowing optimal adjust-ment between fodder supply and uptake by fish.General sampling proceduresTo investigate the potential environmental effects of fishfarming, we examined a set of parameters relative to thewater column (i.e., concentrations of four inorganicdissolved nutrients and heterotrophic bacteria) and a setof parameters relative to the bottom (i.e., concentrationof benthic pigments and organic matter in sediment,size-grain structure of sediment, and abundance of benthic macroinvertebrates). We evaluated the impactlevel by comparing parameter values between samplesfrom fish farms and samples from control zones. Eachfarm had a control zone located about 3 nautical milesaway from the farm and in the opposite direction to theprevailing current in the area in order to minimize poten-tial interactions with dispersed farm wastes. We selectedcontrol zones on a bottom as similar as possible to that of the corresponding fish farm, also characterized by similardepth (±3 m) and distance from shore (±80 m).Samples were collected bi-monthly in March, May,and July 2003, capturing the transition between pre-summer and summer environmental conditions, which islikely the most dynamic and complex environmentalsituation in the Mediterranean coastal system. Here, weassumed that sampling all year round is not necessary, Fig. 1  Maps showing thelocation of the studied fishfarms along the SpanishMediterranean coast122  because if crucial environmental and biotic parametersare affected by fish farming, differences between controlsand farms should be detectable at any time. Samplingwas conducted by divers early in the morning before fishfeeding. Samples were immediately sent by refrigeratedcourier to the Centro de Estudios Avanzados de Blanes(CEAB-CSIC), where we carried out the analyses within36 h of sampling.Dissolved nutrients and bacterioplanktonWe investigated total concentrations of four inorganicdissolved nutrients: orthosilicic acid (hereafter referredto as ‘‘silicate’’), nitrate, nitrite, and phosphate. Al-though silicate does not occur in farm wastes, weinvestigated silicate levels around the farms to evaluatethe possibility that inputs of nitrate and phosphate fromfarms in areas with elevated dissolved silicate concen-trations of natural srcin could fuel diatom blooms. Inthis approach we failed to measure nitrogen compoundsderived from fish excretion directly, which consists of upto 85–90% ammonia (NH 4+ ) and 5–10% urea (Dosdat2001). After excretion, ammonia is quickly oxidized tonitrate, and urea transformed in ammonium, which israpidly taken by both phytoplankton and bacteria.Therefore, although we did not measure nitrogen com-pounds from fish excretion, we evaluated nitrate valuesand the potential effects of excess ammonium in bacte-rioplankton and benthic chlorophyll. We discardedmeasurements of planktonic chlorophyll because thisparameter has been shown to be a bad descriptor of fish-farm impact due to rapid water replacement around thecages (Pitta et al. 1999), a circumstance that particularlyapplies to our study of semi-exposed farms.By using 250-ml polypropylene bottles, we sampledseawater ( n =5) for nutrient determination in fish farms1 mbelowtheoceansurface(surfacewater)and1 mabovetheseafloor(bottomwater).Wemeasuredconcentrationof nutrients immediately upon the arrival of samples at thelaboratory using a TRAACS-2000 Autoanalyzer.We investigated concentrations of heterotrophicbacteria by subsampling 3 ml seawater from the poly-propylene bottles. Subsamples of surface and bottomseawater from each fish farm and control ( n =5) weredelivered to 1.5 ml criovials, fixed in 10% paraformal-dehyde (1%) plus glutaraldehyde (0.05%) at roomtemperature in the dark for 10 min (Marie et al. 1996).Samples were then stored at   80  C until bacterialcounting, which took placed within 2 weeks. We used aFACScalibur flow cytometer emitting at 488 nm, inwhich bacteria are detected by their signature in a sidescatter versus green fluorescence plot, after defrostingand fluorescent staining with 1.6–5  l M Syto 13(Molecular Probes) for 15 min in the dark (Del Giorgioet al. 1996; Gasol and Del Giorgio 2000). Count cali- bration was provided by adding 5% of a 10 6 ml  1 solution of yellow-green 0.92 Polyscience latex beads towater samples. Counts were based on sample runs at lowspeed (approximately 60  l l min  1 ) and data stored in alog model for 2 min or until 1,000 events had been ac-quired. All samples were processed undiluted.For each fish farm, we examined differences in con-centrations of nutrients and bacteria as a function of ‘‘zone’’ factor (farm zone versus control zone) and‘‘depth’’ factor (surface water versus bottom water)using two-way ANOVA. When data did not meet theassumptions for the ANOVA, we applied appropriatetransformations, as indicated in the graphs.Given that sampling size is moderate ( n =5), weavoided three-factor designs, discarding less relevantfactors that would complicate the statistical approachand weaken the power to detect effects of more relevantfactors. For instance, examination of between-farmdifferences in nutrients and bacteria would be of littleinterest in terms of impact evaluation, because putativebetween-farm differences likely reflect geographical dis-tance between farms. Similarly, examination of differ-ences between sampling times would be of minimuminterest in terms of impact detection, because nutrientsand bacterioplankton are known to show marked sea-sonal changes in the Mediterranean. Consequently, weobviated between-farm variability, also statisticallyanalyzing samples from March, May, and July sepa-rately. Nevertheless, for instructive purposes, the resultsare presented in graphs, the format of which allows easyinspection of between-farm and between-time variability.When significant differences in nutrient or bacteriaconcentration were detected by the two-way ANOVA,we run Student–Newman–Keuls (SNK) ‘‘a posteriori’’tests to identify the groups responsible for the differ-ences. Because we are particularly interested in exam-ining differences between farm zones and their respectivecontrols, and because SNK tests based on small samplesize ( n =5) may have limited power to detect significantdifferences,wespecificallyre-analyzeddifferencesbetween Table 1  Summary of farm features, indicating depth of sea floor below the cages, area occupied by cages at the ocean surface, number of cages, number of years of activity, fodder consumed in 2002, and fish biomass averaged for 2002Farm Depth (m) Surface (m 2 ) Number of cages Opening year Fodder consumed (tons) Fish biomass (tons)No. 1 37 5,103 18 1997 915 394.7No. 2 34 5,890 12 2001 114 42.9No. 3 30 4,252 16 1998 1,091.3 549.6No. 4 37 2,945 6 1995 417.7 138.1No. 5 21 3,926 8 2001 116.7 40.4123  each individual farm and control by time and water type(surface water versus bottom water) using either theparametric  t -test or the non-parametric Mann–Whitneyrank sum test. The results of both pairwise analyses wereconsulted before concluding that significant differencesbetween farm and control sites had occurred.Pigments and organic matter in sedimentWe investigated the content of pigments and organicmatter in the 0.5 cm-thick upper layer of sediment,which was collected by divers drawing 50 ml polystyreneFalcon tubes at the seabed.For pigment analyses, 3-g sediment sub-samples (wetweight) were extracted in 90% acetone. Extractionconsisted of three steps, intercalated with 5-min cen-trifugations at 2,000 rpm. After each centrifugation, thesupernatant was collected using a syringe and subse-quently filtered through a 2  l m-pore filter. All threesupernatants of each sample were mixed and centrifugedat 3,500 rpm for 5 min to minimize turbidity prior tospectrophotometer reads. Sample absorbance was mea-sured using a SHIMADZU UV-2100 Spectrophotome-ter at 750 and 665 nm before and after acidification with0.1 N of HCl, respectively. By using Lorenzen’s equa-tions with the appropriate corrections for sedimentssamples (Lorenzen 1966), absorbances were respectivelytransformed into concentration of chlorophyll  a , whichprovides information about the abundance of benthicmicroalgae, and phaeophytin. The latter is a breakdownproduct of chlorophyll, and the ratio of chlorophyll tophaeophytin is used to indicate the health of the mic-roalgal assemblage.For the analysis of organic matter, we used 3-g (dryweight) sediment sub-samples. After homogenizationand drying at 60  C to constant weight, sediment wascombusted in a Muffle furnace at 500  C for 5 h. Theamount of organic matter in a sample was estimated asthe difference between dry and ash weight.Weexamineddifferencesinchlorophyll a ,phaeophytin,andorganicmatterasafunctionofzone(farmzoneversuscontrol zone) and farm factor (farm no. 1 – no. 5) using atwo-way ANOVA ( n =5). When data did not meet theassumptions for the ANOVA, we applied appropriatetransformations, as indicated in the graphic results. Whensignificant differences were detected by the ANOVA, weran SNK tests to identify the groups responsible for thedifferences, paying special attention to pairwise compari-sons between each farm and its corresponding control.Samples from March, May, and July were analyzed sep-arately to avoid a low-power three-factor approach.Granulometric structure of sedimentWe investigated the grain-size structure of sedimentusing 100 g (dry weight) samples collected by a diver-operated hand corer. Formaldehyde-fixed samples wereoxidized in 6% hydrogen peroxide for 48 h and dried at60  C to constant weight. Subsequently, we processedsediment using an electrical CISA sieve, which separatessediment into eight grain-size groups (i.e.,  x <0.06 mm,0.06 mm< x <0.12 mm,0.12 mm< x <0.25 mm,0.25 mm< x <0.35 mm, 0.35 mm< x <0.5 mm, 0.5 mm< x <0.75 mm, 0.75 mm< x <1 mm,  x ‡ 1 mm).Because dredging and the addition of foreign sand forbeach restoration is a common activity at some of thestudied localities from March to June, we determinedsediment structure in March (when a natural post-wintersituation is expected) and July (when an artificial pre-summer situation is expected). Granulometric structurewas expressed as weight percentage of the various grain-size fractions, and fractions named according to theUdden–Wentworth scale (Wentworth 1972).Additionally, we estimated quantitatively the simi-larity among the soft bottom of the diverse farms andcontrol zones using classification analysis. The analysiswas based on the untransformed average (March andJuly samples) of weight percentage for each of the eightgrain-size fractions plus chlorophyll  a . We calculatedpairwise granulometric distances between all 10 zonesusing the Bray-Curtis distance (BCd), which is semi-metric and does not consider the shared absence of descriptors (Legendre and Legendre 1998). For a moreintuitive interpretation, BCd can easily be converted intoits complement, the semimetric Steinhaus similarity(Ss=1  BCd). Finally, the BCd matrix was processed bythe UPGMA clustering algorithm to produce a clado-gram of zones based on ‘‘sediment distances’’.Macroinvertebrate faunaFor the faunal analysis, we collected one hand-corer(10 cm · 15 cm) by Scuba diving at each zone in March,May, and July. Sediment samples were fixed in 4%formaldehyde and sieved through a 500  l m-mesh sieveto separate the macroinvertebrates. The organisms werepreserved in 70% ethanol, stained with 1% Bengal Rose,then counted and identified under dissecting and com-pound microscopes. Animals were classified to thefamily level, an approach that provides an efficient‘‘research effort/result resolution’’ ratio when assessingthe environmental impact of fish farming (Karakassisand Hatziyanni 2000).To increase test power in detecting faunal differencesbetween control and farm samples, we pooled samplescollected from March, May, and July, disregarding theexamination of differences as a function of ‘‘time’’ fac-tor. Then, we investigated faunal differences betweencontrols and farms at three levels of complexity. First,we examined differences in taxonomic richness anddiversity between farm and control zones. Subsequently,we used cluster analysis to explore faunal differencesbetween a given farm and its control in comparison tothose between each of the remaining farms and theirrespective controls. Finally, we investigated the level at 124  which the environmental variables under study (i.e.,dissolved nutrients, bacterial concentration, grain-sizestructure of sediment, benthic chlorophyll  a , and organicmatter in sediment) are responsible for differences infaunal distribution between control and farm zonesusing both unconstrained and canonical ordinationanalyses.Taxa richness was estimated as number of familiesrepresented in each zone. Biodiversity was estimated bythe unbiased Simpson’s Diversity Index ( D ), as proposedby Pielou (1969). The value of   D  ranges between 0 and 1;the greater the values, the greater the sample diversity.To examine differences in  D  values between farm andcontrol zones, we used a paired  t -test, after checkingdata for normality and homoscedasticity. We used apaired test rather than a regular (non-paired)  t -test, be-cause diversity values for each farm and its control arenon-independent statistically, i.e., are associated becauseof geographical closeness.At a second level of analysis, we examined quanti-tative faunal differences between a given farm and itsrespective control by calculating pairwise BCd. Subse-quently, the BCd matrix was processed by the UPGMAclustering algorithm to produce a cladogram of faunaldistance among zones. In this analysis, we considered alltaxa collected in the study and analyzed their abun-dances as untransformed data.Finally, we assessed the explainable faunal variationin the ‘‘zone per taxa’’ matrix using unconstrained cor-respondence analysis (CA). Then, we used canonicalcorrespondence analysis (CCA) to estimate the portionof variation related to the environmental variablesmeasured in this study. Because rare taxa may producedistortion of ordination scores, we excluded taxa withabundance values <3 from these analyses. In addition,we down-weighted rare species, using the option avail-able in the CANOCO 4.0 software. Given that impactedcommunities are often characterized by dramaticallyuneven distributions of abundance values across taxaand sites, we ran the analyses on untransformed abun-dances rather than on log-transformed ones to maximizethe chances of detecting anomalies in abundance dis-tribution.In CCA analyses, we initially considered a total of seven variables; three of them related to the benthiccomponent (i.e., average percentage of chlorophyll  a ,organic matter in sediment, and mud in sediment), thefour others to the water column (i.e., average concen-tration of phosphate, nitrite, nitrate, and bacteria inbottom water). Concentration of silica was not consid-ered in these analyses because it is not really fish-farmwaste. The relevance of the environmental variables tobe considered in the analyses was checked by pre-liminary ‘‘manual addition’’ tests, as well as by inspec-tion of the variable inflation factors (VIF) of trial models.These exploratory CCAs revealed that, although theglobal model based on these seven variables is statisticallysignificant, ‘‘bacteria’’ and ‘‘organic matter’’ are redun-dant variables, explaining a portion of variation alsoexplained by the ‘‘mud’’ variable. Therefore, we ran asecond five-variable analysis considering just three water-column variables (i.e., phosphate, nitrate, nitrite) and twobenthic variables (chlorophyll  a  and mud).In the five-variable approach, we examined theenvironmental sources of variation within the ‘‘taxa perzones’’ matrix, discriminating the variation uniquelydescribed by each set of variables (i.e., water-columnversus benthic ones), that jointly explained by both sets,and that unexplained by the analysis (Borcard et al.1992). The statistical significance of the first canonicalaxis and all canonical axes of the resulting models weretested by the Monte-Carlo test using 500 permutationsunder the reduced model or under the full model if co-variables were considered in the analysis.To generate bi-dimensional ordination diagrams, inwhich zones and taxa are represented by points andquantitative environmental variables by vectors (inCCA), we opted for a biplot scaling focused on inter-zone distances. Results Dissolved nutrients and bacterioplankton Silicate The most obvious global trend is that the concentrationof silicate progressively decreases from spring to summerin both control and fish-farm zones, also in both surfaceand bottom waters (Fig. 2a). This is likely the result of the spring bloom characterizing the populations of western-Mediterranean diatoms, which are major sili-cate consumers. Despite major global patterns, a moredetailed analysis of silicate concentration for each fishfarm indicates a disparity of trends as a function of zoneand depth factors, depending on fish farm. It is note-worthy that zone factor was statistically significant inonly five out of the 15 analyses, while depth factor wassignificant in nine analyses. More importantly, in thosecases in which zone factor was significant, the SNK testsindicate that the sign of the difference between farmwater and control water is not consistent; sometimesconcentration is higher in farms and sometimes in con-trols, depending on farm, depth and time. Phosphate Phosphate concentration remained consistently low overthe study period, in both control and fish-farm areas,also in surface and bottom waters (Fig. 2b). This isconsistent with the phosphorous-limited nature of thewestern Mediterranean (Hargrave et al. 1991). We alsonoticed that the sign of differences in phosphate con-centration as a function of zone and depth shifteddepending on fish farm and sampling time. In all cases,differences were very small in magnitude. Although the 125
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x