3. Application of methodology and data description
The project used a combination of methodological approaches to achieve the specific objectives noted previously.
Socio-economic and environmental information was gathered on sea lice control measures employed in the salmon sector based on a review of secondary sources of information from Scotland as well as other countries such as Norway and Canada. Where no Scottish data were available and data from other countries had to be used, these were inspected and selected using expert opinion (health practitioners in the Scottish salmon sector through telephone interviews and email communications, and members of the research group conducting the study) to ensure closest possible relevance to the Scottish salmon sector, and sensitivity analysis was performed. Secondary data collection was combined with primary data collection using a participatory workshop and in-depth interviews with Scottish salmon producers and processors. The questionnaires used for primary data collection are presented in the Appendix together with the description of the data collection, storage and use rules followed in the project (ethical aspects of data collection/storage/analysis/reporting). Quantitative and qualitative data collected through survey questionnaires was used to inform the modelling and reference is made to it as part of modelling assumptions and data description.
Results from the participatory workshop with stakeholders representing different stages of the supply chain was used to inform our analysis of farm-level behaviour (uptake of control measures). To run this workshop, we employed a newly developed participatory process from the system dynamics literature called group model building (GMB) (Rich et al. 2018) where stakeholders rank control measures based on their efficacy (estimated and/or perceived), and then collectively identify incentives linked to different stages of the supply chain to reduce occurrence of sea lice in primary production. The involvement of all stakeholders contributed to identifying network or spill-over effects, supply chain constraints and gaps in skills and training requirements. Methodology is described in the section presenting results of the workshop.
Cost Effectiveness Analysis (CEA) and Life Cycle Analysis (LCA) were used to assess the relative cost-effectiveness of sea lice management measures and their impact on the economic performance (including carbon cost) of the Scottish farmed salmon industry. CEA and LCA have not been used, to the best of our knowledge, in other studies focussed specifically on the control of sea lice in the salmon sector; however they have been extensively used in studies analysing the economic and environmental impacts of control of animal disease.
Cost-effectiveness analysis (CEA) is a technique used to assess measures to prevent or mitigate a disease where the impact cannot be measured routinely in monetary terms, which we used in this context to evaluate different sea lice management measures. Many researchers working on animal and human health economics have used this method to determine the optimal resource allocation between interventions at their disposal (Mangen et al., 2007; Valeeva et al., 2007; Benedictus et al., 2009). The Cost-Effectiveness ratio (CE ratio) is a commonly used indicator to determine the effectiveness of an intervention (Rushton and Jones, 2018). This ratio is represented as:
Determining the cost of the intervention and effectiveness of the intervention are essential steps to determine applicable CE ratios. The cost of the intervention includes all costs of controlling or preventing a disease. It may consist of the cost of licenced veterinary medicines, equipment, implementing procedure including means of application, labour costs, administrative costs etc. The effectiveness of an intervention is used to compare how effective different prevention or mitigation measures are, and is generally represented by an efficacy score for each measure. This score is determined based on the specific aim of a study. For instance, Bergevoet et al. (2009) used lowering prevalence of salmonella in dairy herds, and Valeeva et al. (2007) used improvement in food safety on dairy farms as efficacy scores to conduct their cost-effectiveness assessment studies. Our study measured the relative cost-effectiveness of sea lice management measures for farmed salmon production (methodology details are presented below).
The Life Cycle Assessment (LCA) method considers the environmental burdens and resource use in the production and exploitation of a commodity within defined boundaries. The commodity (or end product) considered in the analysis is called Functional Unit in the LCA terminology, and it must be clearly specified and consistently used (also in terms of quantity) throughout the assessment. The boundary can be from cradle to grave, which includes the production, retail, consumption, and disposal stages, but it is also common, for pragmatic purposes, to focus the analysis on the earlier stages, for example in agricultural production, this may include stages up to the farm gate. LCA can be considered to be the most holistic method available for environmental impact assessment, and therefore it is the methodology favoured by major organisations, such as the United Nations Environment Program.
The methodological framework used in this study combines LCA and CEA, with the former feeding into the latter. We modelled sea lice management measures separately as single use measures. In addition, we modelled three combinations of measures that represent examples of the many realistic sequences of measures possible between stocking and harvesting on a salmon farm. They simulate a situation where producers monitor fluctuating sea lice levels and use management measures as effectively as possible with the aim of maintaining sea lice counts within acceptable (regulation compliant) levels. Results of the LCA and CEA models run for single use measures provide a comparison and ranking between single measures, while those run for combination measures provide examples of total environmental impacts and total costs.
Life Cycle Analysis: modelling environmental impacts (carbon costs)
The LCA model applied in this study covered the whole salmon production chain from raw materials to the farm gate (cradle to gate). The functional unit of the study was one marketable salmon weighting 5 kg at the farm gate. The main components of the modelled system are shown in Figure 1.
Different sea lice management measures have effects on different parts of the production chain. This is also demonstrated in Figure 1. As a result, one measure may have a direct effect on a single component of the chain, but then the changes in this component may cause changes in other parts of the chain. Therefore, each measure could have potential indirect effects on the whole chain, and this can also generate both direct and indirect environmental impacts. To understand all these interactions, a systems approach to LCA modelling is necessary. It should also be noted that if any measures affect the performance of the fish (e.g. growth rate, feed efficiency, mortality and output of marketable fish), this is further reflected in the resource efficiency of the production through indirect environmental impacts.
LCA: Data sources, assumptions and limitations
LCA is a data-intensive method, and the initial expectations were to base the model as much as possible on primary data collected through interviews with the Scottish salmon sector. Ultimately, the LCA model used a combination of primary and secondary data, and modelling assumptions:
1. Secondary data from literature and publicly available databases. Whenever possible, Scottish and UK data sources were used; non-UK data were selected to relate as closely as possible to the Scottish production and inputs e.g. comparable to other input data sourced from Scotland.
2. Modelling based on functional relationships
3. Expert opinion (health practitioners in the Scottish salmon sector and members of the research group conducting the study)
The limited amount of data on sea lice management measures in the Scottish salmon primary production that was available for this study made it necessary to simplify the modelling approach. This is presented in Figure 2. We assumed that the measures had no effect on the feed conversion ratio (FCR) per se, but the measure-related mortality increased the feed consumption per one marketable 5 kg fish to account for losses from feed consumed by fish not harvested. We assumed that the combination measures were effective in maintaining sea lice within acceptable levels between stocking and harvesting. This meant that, other than the measure-related mortalities, there were no effects of measures on the environmental costs related to production through changes in performance of the fish. We assumed no measure-related resistance as not within the scope of this study and adding a level of complexity not handled without experimental data.
The data for modelling the general performance of the salmon production chain (excluding sea lice management measures) were obtained from a review of the literature (e.g. Newton and Little 2018, Philips et al. 2019, Boissy et al. 2011, Marine Scotland 2018, ecoinvent 2020). The data sources for each separate sea lice management measure are presented in Table 1.
|Sea lice management measure||Inputs||Sources|
|Fresh water (well boat & tarpaulin)||
|Licenced veterinary medicines† (well boat & tarpaulin)||
|Hydrogen peroxide (well boat & tarpaulin)||
|Hydrolicer and Thermolicer and Optilicer||
|Cleaner fish (captured & farmed)||
†includes AlphaMax (deltamethrin), Salmosan Vet (azamethiphos).
Combination sea lice management measures. We model three combination measures (Figures 3, 4, 5). The costs of equipment, implementation and side effects were assumed to be additive under each combination of measures (no resistance).
Source: Mowi, 2018
Source: based on survey data
Source: own assumptions
Sensitivity analysis was performed to account for uncertainty of data inputs for the LCA model. In this analysis, a two-fold amount of all inputs (e.g. diesel, veterinary medicines, licenced veterinary medicines used in the bath measure, fresh water, construction costs of boats and other equipment, cleaner fish, cleaner fish feed) compared to the default values was used, and the resulting greenhouse gas (GHG) emissions for management measures, separate and in combination, were compared with the original results. Sensitivity analysis for combination measures was also run in a purely hypothetical scenario, ‘Combination 4’, with a limited number of sea lice management measures (1 x skirts, 1 x H2O2, 1 x hydrolicer), and a two-fold increase in background mortality/rejects, i.e. the standard background mortality (not including any additional mortality caused directly by the sea lice treatments) from 15% (assumed in this study) to 30%. This part of the sensitivity analysis was performed to demonstrate the impact of potential output losses on the carbon costs of the production chain.
Cost Effectiveness Analysis
We aimed to determine cost-effectiveness of different sea lice management measures on Scottish salmon farms. We considered the reduction in sea lice count as an indicator of effectiveness used for this study. This was not based on quantitative evidence but on a combination of expert opinion (aquaculture experts and health practitioners in the Scottish salmon sector), and literature review as the effectiveness measure.
We identified the cost of intervention under each of the sea lice management measures into 4 categories; cost of equipment, cost of implementation, environmental cost and cost of side effects (Figure 6). The accumulative cost was assessed against the reduction of sea lice on an adult salmon fish, which is considered as the efficacy score for this study (not experimentally assessed). The updated CE ratio used in this study is:
Where, Tn is the nth measure, Ceqp is the cost of equipment, Cimp is the cost of implementation, Cenv is the environmental cost, Cse is the cost of side effects and EF is the efficacy score
i. Cost of equipment, Ceqp:
This cost includes the cost of licenced veterinary medicines (if any) used and other especial equipment required to use the sea lice management measures. For instance, for H2O2, this cost includes cost of H2O2 and cost of oxygenation.
ii. Cost of implementation, Cimp:
This covers the cost of the application of the sea lice management measures. It includes costs of boats, labour and other provisions that are required to implement the measures. For example, to implement H2O2, this cost will include well boat and labour.
iii. Environmental cost, Cenv:
The carbon cost for applying each of the sea lice management measures is determined by using a Life Cycle Analysis (LCA) model developed for this study. The LCA uses carbon emissions from all the activities associated to providing and implementing each measure. The emissions are converted into carbon cost by using a standard carbon cost rate £12.8/tCO2eqv.
iv. Cost of side effects, Cse:
This cost includes any loss in revenue due to increased mortality of salmon under a sea lice management measure.
Where, m is increased mortality under sea lice management measure Tn, ps is price of salmon, fs is feed saved due to starvation and pf is price of feed per kg
The first limitation of this study was the limited availability of secondary data sources and limited primary data or access to it. As a result, the majority of the financial information related to different sea lice management measures applied on Scottish farms were taken only from one source (Macaskill, 2014). For information not available for Scottish farms, we used estimates based on similar information available from other salmon producing countries, such as Norway (Iversen et al., 2017). As mentioned in relation to LCA limitations, we have selected data as closely as possible to the Scotland/UK salmon sector, and detailed sensitivity analysis was performed.
A second limitation was the absence of control farms to compare our results with. Most health economic studies rely on empirical data in a ‘with- and without- ‘format which provides a reliable source of generating the indicators. Sea lice infestation is a significant problem for Scottish salmon producers so a ‘without-sea lice management measure’ scenario is almost non-existing. For this reason, we have used expert opinion on efficacy of each of the single use measures to obtain qualitative relative efficacy scores. Because there was a wide variability in the expert opinion on the efficacy of sea lice management measures, we included a sensitivity analysis to explore how the minimum and maximum efficacy estimates affect model outcomes.
Another limitation of this study concerns the combinations of measures. Because there is no standard combination of sea lice management measures that farms use, we modelled single use measures separately, and three realistic combination measures based on expert opinion (survey data) and members of the research team (as it was beyond the scope of this study to include all possible combinations).
We used the following assumptions relating to a Scottish salmon farm and sea lice management measures (Table 2). We assumed that an average salmon farm has 1.2 million fish with an average harvest weight of 5 kg per fish. The average length of the production cycle is 20 months in the marine environment and each fish required 5.75 kg of feed per production cycle, when assuming a 15% “background” mortality. An average farm gate price of live salmon is set to £32.42 and average price per kg of feed is £0.94. For the cleaner fish, it was assumed that 48,000 fish are required on an average salmon farm. We also assumed that synchronised fallowing is the standard practice on each Scottish salmon farm and hence, it was not included as a sea lice mitigation method in this study.
|Fish||1.2 million per farm||Macaskill (2014)|
|Harvest weight||5 kg|
|Length of production cycle||20 months|
|Feed requirement||5.75 kg per fish per cycle|
|Salmon farm gate price||£32.42 per fish|
|Cleaner fish requirement||48,000 per farm||Macaskill (2014)|
Considering the large variation in efficacy of a sea lice management measure, we assumed the efficacy score of each measure was consistent and based on a general optimal outcome. We also assumed costings of each single use measure to be additive when used in combination and achieve the efficacy necessary in bringing down and maintaining sea lice count within acceptable (regulation compliant) levels i.e. with no impact on fish health, welfare and productivity, thereby ignoring resistance effects or variation in efficacy.
CEA: Costs of equipment, implementation and side effects
The sea lice management measures included in this study could be grouped based on characteristics. The costs of equipment, implementation and side effects associated with these groups of measurements are briefly described below.
i. Sea lice management by fresh water bath can be implemented in two ways – using tarpaulins or using well boats. The costs of fresh water batch treatment were provided for both implementation procedures as shown in Table 3 and Table 4.
Fresh water using well boat: For this measure, the costs included were water costs, cost of a well boat and labour costs.
|Costs (£/per measure)||comment|
|Water||1008||Water charge £0.13/m3*; well boat capacity 2500m3**|
|Well boat||18,000||Macaskill, 2014|
Fresh water using tarpaulin: the costs included water costs, tarpaulin and labour costs.
|Costs (£ per measure)||comment|
|Water||1008||Water charge £0.13/m*|
The bath medicines included hydrogen peroxide, AlphaMax (deltamethrin), Salmosan Vet (azamethiphos). Baths can be implemented in two ways – using tarpaulins or using well boats. The costs for both implementation procedures were as shown in Table 5 and Table 6. Due to lack of information on individual prices as currently used in the Scottish salmon industry, cost of licenced veterinary medicines were taken from literature.
|Costs (£ per measure)||comment|
|Licenced veterinary medicines||30,000||Macaskill, 2014|
|Costs (£ per measure)||comment|
|Licenced veterinary medicines||30,000||Macaskill, 2014|
iii. In-feed (Slice):
The cost of in-feed measures included licenced veterinary medicines, equipment and implementation costs. No separated costs were available and hence, total cost was determined to be £35,000 per measure based on Macaskill, 2014.
iv. Physical removal:
The physical removal measures currently relevant are hydrolicer, thermolicer and optilicer. Costs for these measures were not available for Scottish salmon farms and hence were based on information provided in a project report assessing the main production costs in Norwegian salmon farms (Table 7). It should be noted that the physical removal measures are only applied on farms to adult salmon fish (> 1 kg of weight). For smaller and younger fish, other sea lice management measures are used.
|Total Costs per measure||comments|
|Hydrolicer||139,104||Includes machine depreciation, cost capital, service vessels, fuel and labour|
|Thermolicer/Optilicer||181,440||Includes machine depreciation, cost capital, service vessels, fuel and labour|
Source: Iversen et al., 2017 http://hdl.handle.net/11250/2481501
The skirts are tarpaulins with an open bottom. They form physical barriers around salmon pens to minimise contact between sea lice drifting in from outside the pen, and fish. There are different kinds of materials (dense or semi permeable) used to make skirts. We assumed that skirts used in Scottish farms were dense skirts (e.g. tarpaulin) which generally could survive at least the whole salmon production cycle. We used total costing of applying skirts (£48,384 per cycle) which included cost of the material, application and maintenance of the skirts.
vi. Biological (Cleaner fish)
In Scotland, the common cleaner fish used are different species of wrasse and lumpfish. These are permanently employed in the salmon pens. The costings used in the model were for wrasse species as shown in Table 8.
|Costs (£ per measure)||comment|
|Fish (wrasse)||91,200||£1.9 per fish; 48000 fish|
|Hides||11,040||£230 each; 48 hides|
|Feeders||2,400||£50 each; 48 feeders|
Source: Macaskill (2014)
CEA: Environmental costs
The environment costs of each of the sea lice management measures are presented in Table 9. These are technical inputs used in the LCA model, and not comparable with each other as based on different functional units. CO2e emissions generated by the LCA model are presented in Table 10. These costings were based on emissions from the whole production cycle when using one sea lice management measure. A standard carbon cost of £12.8 per tCO2e was used to determine total carbon emissions per fish (BEIS, 2019).
|Sea lice management measures||Carbon costs (£ per fish)||Mortalities♯ (% per measure)||Carbon costs (£ per fish) in sensitivity analysis when assuming 2x input quantities|
|Fresh water (well boat)||0.0012||0.5||0.0018|
|Fresh water (tarpaulin)||0.0006||0.5||0.0010|
|Hydrogen peroxide (well boat)||0.0020||1.0||0.0029|
|Hydrogen peroxide (tarpaulin)||0.0016||1.0||0.0020|
|Other licenced veterinary medicines † (well boat)||0.0010||0.5||0.0014|
|Other licenced veterinary medicines † (tarpaulin)||0.0008||0.5||0.0010|
|Cleaner fish (capture & farmed)||0.0003||0.00||0.0006|
|Entire production cycle|
†includes AlphaMax (deltamethrin), Salmosan Vet (Azamethiphos).
♯Source: for mortality of physical removal measures (Iversen et al., 2017 http://hdl.handle.net/11250/2481501); for others (expert opinion)
Source: own calculation (LCA)
|Sea lice management measures||Total lifecycle GHG, t CO2e per harvested 1 t liveweight at farm gate at the end of the cycle||Total lifecycle GHG, t CO2e per harvested 5 kg fish at farm gate at the end of the cycle||Total lifecycle GHG, t CO2e per harvested 5 kg fish at farm gate at the end of the cycle in sensitivity analysis when assuming 2X input quantities|
|Combination 4 (sensitivity analysis assuming 2X mortality)||2.377||0.01189||0.01196|
The results with the default inputs and additional results from the sensitivity analysis show that the differences in the GHG emissions between the different measures are small, and due to the high uncertainty in the input values, any ranking or detailed comparison between the measures is not meaningful. Inclusion of the hypothetical ‘Combination 4’ scenario shows that a high increase in mortality/rejects could potentially result in a strong increase in the GHG emissions, and this increase could be much higher than the direct emissions related to any sea lice management measures. The main reason is that the increased mortality would reduce the number of harvested fish, while a large part of the resources (most notable feed) would be wasted, and therefore the resource use would be higher per harvested fish (e.g. higher FCR). In such a case, the sea lice measures should not be considered as a source of carbon costs, but instead as a method to achieve carbon saving as a result of more efficient production.
CEA: Efficacy scores
The efficacy scores of sea lice management measures represent the relative efficacy and are based on expert opinion. Experts were asked to base their perceived efficacy on the difference between pre- and post- sea lice counts around a measure through a questionnaire. Expert opinion considered in this study consisted of health practitioners and experts in the Scottish salmon sector who participated in the workshop (three out of ten attendees provided further responses) or responded to the in-depth interview, members of the research group conducting the study, and expert opinion recorded in the literature (SAIC, 2019). The physical removal measures were the highest-ranked measures achieving more than 80% of efficacy scores and hydrogen peroxide was the lowest-ranked measure with 43% efficacy score (Table 11). The sensitivity analysis included maximum and minimum levels of efficacy scores for each sea lice management measure.
|Sea lice management measures||Efficacy scores||Efficacy range|
|Fresh water (well boat & tarpaulin)||0.64||1.00||0.20|
|licenced veterinary medicines† (well boat & tarpaulin)||0.60||0.90||0.50|
|Hydrogen peroxide (well boat & tarpaulin)||0.43||0.60||0.10|
|Cleaner fish (capture & farmed)||0.72||0.90||0.60|
†includes AlphaMax (Deltamethrin), Salmosan Vet (Azamethiphos).