MODELLING THE IMPACT OF FINFISH AQUACULTURE ON SEDIMENT BIOGEOCHEMISTRY

Transcription

1 EU FP6 ECASA project MODELLING THE IMPACT OF FINFISH AQUACULTURE ON SEDIMENT BIOGEOCHEMISTRY Daniele Brigolin, University of Venice Chris Cromey, T.D. Nickell SAMS, Oban, Scotland (UK) D.R. Aguilera, Pierre Regnier University of Utrecht, The Netherlands Pastres Roberto University of Venice, Italy ECEM 27, Trieste November 27-3

2 Outline Impacts of finfish farming on the sediment geochemistry; Description of the model; Study site and field data description; Results of model calibration; Response of model output to the establishment of a new salmon farm; Perspectives.

3 Finfish farming impacts on the sediment Changes in the chemical properties of the sediments provide a measure of the impact caused by the fish farm which is integrated over time. Wasted food Faeces currents sedimentation S=, NH4+, DIP burial Potential impacts on the sediment (e.g. Hargrave et al., 1997; Pearson & Black, 2) Increase in the superficial OC concentration Decrease of the oxygen penetration depth Increase in dissolved nutrient concentrations Increase in S= concentrations Changes in macrofaunal community Oxic and anoxic mineralization Benthic impacts are proportional to: Exposure of the site (currents); Changes in the fluxes at the sediment-water interface Husbandry practices (feeding strategy and quality of the food)

4 Modelling the impacts of finfish farming on the sediments Current approach OC flux reaching the sediment-water interface Biotic indexes (ITI, AMBI) Approach proposed in this work Simple geochemical indicators (TOC) OC flux reaching the sediment-water interface More complex geochemical indicators (e.g. NH4+, S=, O2..) Ecological indexes

5 Conceptual model: Deposition + Early diagenesis DEPOMOD particle tracking model (Cromey et al., 22 Aquaculture) (validated for Scottish sealochs) Food wastage and Faeces production WATER COLUMN Deposition model (DEPOMODTM) Organic carbon flux SEDIMENT Current velocity O2 demand, nutrient fluxes EDM: Early Diagenetic Model (developed in BRNSTM environment) Sediment temperature

6 Objectives of this work 1) evaluating the applicability of a Reactive-Transport Model of early diagenesis (EDM) in combination with the DEPOMOD; 2) testing DEPOMOD+EDM integrated model at sites which are exposed to high organic carbon fluxes from aquaculture activities, under transient conditions. The potential use of the integrated model is discussed in relation to benthic biogeochemical indicators for costeffective EIA and monitoring practices.

7 Organic matter degradation and early diagenesis processes: conceptual model The modeller specifies the Water column Temp. fluxes at the upper boundary for the solid species, and the concentrations for the dissolved species Fluxes: Concentr.: OC, Fe(III), Mn(IV) O2, NH4+, HPO42-, NO3-, SO42-, Mn2+, Fe2+ (1 ϕ ) C s (1 ϕ )ω Cs (1 ϕ ) DB Cs = + + R ( (1 ϕ ) C s, β, t ) t z z z Surficial sediment layer Deep sediment layer ( ϕ Cw ϕ ω C w ϕ DB C w ϕ Dm / 1 ln ϕ = + + t z z z z 2 )C w + R(ϕ C w, β, t ) t: time z: depth ΣR variation in the concentration due to biogeochemical processes DB diffusion coefficient ω : sedimentation rate φ: sediment porosity. Lower boundary condition: null gradient

8 Organic matter degradation and early diagenesis processes: The modeller specifies the conceptual model Water column Temp. fluxes at the upper boundary for the solid species, and the concentrations for the dissolved species Fluxes: Concentr.: OC, Fe(III), Mn(IV) O2, NH4+, HPO42-, NO3-, SO42-, Mn2+, Fe2+ (1 ϕ ) C s (1 ϕ )ω Cs (1 ϕ ) DB Cs = + + R ( (1 ϕ ) C s, β, t ) t z z z Surficial sediment layer Deep sediment layer ( ϕ Cw ϕ ω C w ϕ DB C w ϕ Dm / 1 ln ϕ = + + t z z z z 2 Operator splitting method z:implicit depth Numerical scheme )C w + R(ϕ C w, β, t ) t: time ΣR variation in the concentration due to biogeochemical processes DB diffusion coefficient Regnier et al. ω : sedimentation rate(22) Appl. Math. Model. φ: sediment porosity. Lower boundary condition: null gradient

9 BRNS Early diagenesis model Reaction network: Primary & Secondary reactions Organic Matter Fluxes of NH4+, O2, DIP Sediment Water Interface Organic matter Degradation redox reactions Reduced compounds Fe2+, Mn2+, NH4+, S= Electron acceptors: O2, NO3-, Fe(III) Mn(IV), SO42- Re-oxidation reactions Electron acceptors O2, Fe(III), Mn(IV) FeS & Carbonates Precipitations Carbonates equilibria

10 Study site Loch Creran, West coast of Scotland The fish-farm was moved to the actual site in Feb 26 A purposely-designed field campaign was carried out in August 26, in order to apply the integrated model

11 Study site Loch Creran, West coast of Scotland 3 stations : 1m and 4m from the cages, and control; 3 replicates per station using a megacorer (Φ =1 mm); Analysis were performed at each 1 cm on the top 1 cm + surnatant waters. Parameters sampled: - Porosity; - Organic Carbon in the sediment; - Fe2+ and Mn2+ in pore waters; - NH4+ and HPO42- in pore waters; - SO42- in pore waters.

12 Application of the model at Loch Creran - methodology Step 1: the Early Diagenesis Model (EDM) was calibrated, by comparing the steady-state model outputs with the field data collected at station BC (control); Step 2: the DEPOMOD was run, in order to obtain a prediction of the farm originated Organic Carbon flux at stations B1 and B4; Step 3: organic carbon fluxes predicted by DEPOMOD are added to the background OC fluxes, moving the EDM from a steady-state to a transient-state; Step 4: transient profiles predicted by the model are compared with a set of sediment chemistry data purposely collected at stations B1 and B4.

13 Step 1, results: EDM calibration station Bc - All the parameters of the model were fixed on the basis of literature references; - The fluxes of solids OC, Fe(III) and Mn(IV) at the upper boundary were calibrated by minimizing a goal function which quantifies the deviation between model predictions and field data SO4 [mmol L-1] e Fe2+ [µ mol L-1] 8 DIP [µ mol L-1] 3 d 2 c NH+4 [µ mol L-1] OC [%] b a 16 2 f Mn2+ [µ mol L-1] 12

14 Step 2: DEPOMOD output organic carbon flux at the S.W.I. Salmon cages Sampling stations g C m-2 yr-1 2 B4 1 B1 m 1 m % of feed waste was assumed; Farm details: At the time of the survey, the fish-farm has been operating for 6 months. Production: 1 tonn y-1 6 x 22m Φ circular net cages reaching 14m depth

15 Step 3: EDM transient simulation Depomod output OC (food+faeces) flux at Station B1 These fluxes were added on the top of the background OC flux, which was estimated by calibrating the EDM. This additional OC flux was imposed for 6 months (age of the farm), perturbing the steady-state profiles and driving the model to a transient-state.

16 Step 4. EDM model output vs field data at station B SO4 [mmol L-1] e Fe2+ [µ mol L-1] 8 DIP [µ mol L-1] 2 3 d 2 c NH+4 [µ mol L-1] OC [%] 1 b 1 a 16 2 f Mn2+ [µ mol L-1] - Due to the mineralization of elevated quantities of fish farm-derived labile organic matter, nutrient concentrations at station B1 are greatly enhanced compared with the non-impacted station BC, reaching values approximately 1 times higher - A sub-surface maximum in nutrient concentration is localized around cm depth

17 Model predicted fluxes at the sediment-water interface NH4 efflux and sediment oxygen demand, are predicted by the model. NH4 efflux mmol m-2 d Bc B1 The model predicts a relevant enhancement in NH4+ efflux at the station localized underneath the farm, with respect to the control station.

18 Perspectives Farm management practices In order to test the integrated model sensitivity to boundary conditions, a scenario was run by assuming a lower percentage of fish food wasted, 2% instead of %, % waste 2% waste % waste S(-II) [mmol L ] % waste NH+4 [µ mol L ] These results indicate that the integrated model may represent an useful tool for studying the level of impact on sediment geochemistry associated with different management practices.

19 Perspectives - Prediction of cost-effective indicators The future aim of this class of models might typically be to focus on the prediction of cost-effective indicators of organic enrichment. Wildish et al. (21) concluded that monitoring programs based on geochemical measures are more cost-effective than the ones based on macrofaunal community measures. The current efforts to provide in situ monitoring of sediment ecosystems (e.g. are improving the suitability of this class of geochemicalbased measures. According to Hargrave et al. (1997) S= concentration and sediment oxygen demand are among the most sensitive indicators of fish-farm organic enrichement S(-II) [mmol L ]

20 Concluding remarks Two numerical models, simulating fish-farm waste deposition and early diagenesis processes in the sediment, were applied at a Fjordic Sealoch, for studying the impacts of finfish aquaculture on sedimentary redox dynamics. The early diagenesis model (EDM) was calibrated by using a set of original field data of selected pore water and solid-state chemical species measured in the field at a pristine site. The EDM response to an increase in OC rain due to the installation of the fish farm was studied by comparing model predictions with field data. The study of the EDM behaviour under transient conditions and the potential use of the two integrated models for aquaculture site-selection and monitoring purposes were the most relevant aspects discussed in this work.

21 The authors gratefully acknowledge the assistance of Ms Susan McKinlay with core slicing, Mrs Heather Orr with CHN analysis, Ms Cheryl Haidon with metals analysis, Mr Tim Brand for phosphate analysis and Mr Martyn Harvey with sulphate analysis Thank you!