Mapping the impacts of farmed Scottish salmon from a life cycle perspective
The European Union relies on seafood imports to supply growing demand that European production has failed to meet. Politically motivated media reports have denigrated competing imports in favour of local production. While life cycle assessment (LCA) measures global impact of value chains, it often fails to contextualise them. Using LCA, this article takes farmed Scottish Atlantic salmon as a case study of “local” production to identify and map the contributions to global environmental impact.
Data on the Scottish salmon value chain were collected by structured survey from a large international feed mill, six farms and a major processor. Secondary data were collected from available literature on feed ingredients and background data from EcoInvent2.2. A mid-point CML2001 approach was adopted focussing on global warming potential (GWP), acidification potential (AP), eutrophication potential, ozone depletion potential, photo-chemical oxidation potential, consumptive water use and land use. Results were displayed as contribution analyses of materials and processes and mapped geographically using area plots.
Results and discussion
Far from being a “locally” produced commodity, nearly 50% of the feed ingredients were sourced from South America and less than 25% originated in the UK. It was found that over 90% of the impact to farm-gate was embodied in feed, apart from eutrophication potential which was high at the farm from direct nitrogenous emissions into the marine environment. The majority of impacts do not occur in Scotland, particularly for land and water use, which occur at a more geographically significant level than GWP or AP, which are more global or regional impacts, respectively. High GWP emissions from vegetable-based ingredients were related to soil management and energy intensive processes such as wet milling to produce gluten from wheat and maize, sunflower and rapeseed oil processing.
The results show that in an age of globalised commodity trading, concerns around “local” production are often misleading. As consumers try to make more responsible purchase choices, they may be misled over the global impacts their choices are having. There are clearly trade-offs between different feed ingredients, especially regarding substitution of marine ingredients with those of vegetable origin. While marine ingredients perform comparatively well, they are highly limited, and biodiversity impacts of different ingredients are less clear and difficult to compare.
KeywordsAtlantic salmon Impact mapping LCA Local production Scotland
The European Union is becoming ever more reliant on seafood imports to supply growing demand that European fisheries and aquaculture have failed to meet (Little et al. 2012). Politically motivated media reports have made claims that competing seafood imports may be associated with environmental and social concerns, and the EU should be seeking to import less, in favour of promoting growth in European aquaculture. This aligns with current consumer interests in consuming “local” food as a lifestyle choice (Janßen and Langen, 2016). However, as trade becomes more globalised, and competition for resources increases, raw materials for livestock production are increasingly sourced on a global basis, which inevitably result in environmental trade-offs (Ziegler et al. 2016).
Apart from smolts (juveniles), Scottish Atlantic salmon (Salmo salar) is produced exclusively in net pen systems in sheltered marine lochs or bays. Production started in the 1970s, increased quickly during the 1980s and 1990s but subsequently plateaued due to production challenges and increased competition for suitable sites with other stakeholders. Recently, production has begun to increase as the Scottish government has set a target to produce over 200,000 t of salmon by 2020 (http://www.gov.scot/Topics/marine/Fish-Shellfish accessed 5/7/2017). Scottish salmon is now exported to over 60 countries worldwide (SSPO 2015) and is the second most valuable UK food export after whisky. In some circumstances, salmon aquaculture has raised environmental concerns from a number of standpoints, both locally and globally.
LCA is an ISO-accredited (ISO 2006a, b) environmental impact accounting tool which measures the cumulative impact throughout a product’s value chain, covering a range of different environmental impacts. Several LCAs of salmon have been conducted in the past, notably Pelletier et al. (2009), Ayer and Tyedmers (2009), Torrisen et al. (2011) and McGrath et al. (2015). Some other LCAs have focussed on salmonid feeds such as Papatryphon et al. (2004) and Boissy et al. (2011). Of all the studies related to salmon, only Boissy et al. (2011) included water consumption and land use, and none of them attempted to map the various impacts throughout the value chain (VC), i.e. the connected and interrelated nodes of an industry that create and add value (Dey et al. 2015). Troell et al. (2014a) identified water footprints from a life cycle perspective as an important issue in the assessment of mariculture systems. In previous LCAs of salmon VCs, the vast majority of environmental impact has been associated with feed ingredient sourcing (e.g. Pelletier and Tyedmers 2007; Pelletier et al. 2009). Indeed, it has been estimated that around a third of agricultural land use may be devoted to providing livestock feed ingredients (Robinson et al. 2011) of which aquaculture represents an estimated 4% (Troell et al. 2014b).
Aquaculture feed formulators source ingredients from a global pool of resources such as marine ingredients from South America, which may represent as much as 40% of the diet by volume. Efforts to replace marine ingredients, which are regarded as being highly limited (Naylor et al. 2009; Ziegler et al. 2016), have resulted in greater substitutions of vegetable-based proteins, especially those derived from soybean which is critical for many European livestock feeds (FAO 2006), including for aquaculture (FAO 2006; Torrisen et al. 2011). The vast majority of European-sourced soybean meals and concentrates are also from South America and have been connected to environmental concerns around deforestation of sensitive habitats for agricultural expansion (FAO 2006; Robinson et al. 2011; Tritsch and Arvor 2016; da Costa et al. 2017). Some vegetable-based feed ingredients have also been shown to be highly energy intensive in their processing compared to the marine ingredients which they substitute for (Boissy et al. 2011; Pelletier et al. 2009). There are clearly trade-offs between different impacts with regard to water, land, energy use and their associated emissions and impacts, but also to how impacts are geographically distributed. This is particularly the case for comparing between marine ingredients with low demands, especially on land and agricultural ingredients with potentially high land and water demands. Many aquaculture LCAs have presented a comprehensive set of environmental impacts but have often lacked contextualisation of these impacts from a local or global perspective and the implication for improvements or decision making, and it can be argued that a lack of regionalised/localised impact assessment is common throughout LCA. Several studies have investigated regionalised impacts through LCA (e.g. Schlich and Fleissner 2005; Boulay et al. 2011; Patterson et al. 2011), and some efforts have been made to combine LCA with GIS methodology (e.g. Geyer et al. 2010; Gasol et al. 2011; Dresen and Jandewerth 2012; Mutel et al. 2012); however, these have not been frequently applied and only a few LCA studies have represented data geographically. Other challenges have arisen in how to measure the impacts of land use change related to the dependence of land required for agricultural expansion (Mila i Canals et al. 2007). Different approaches are being developed to measure the impacts of land use change (LUC) such as UNEP-SETAC initiatives, although consensus on regional factors (Nordborg et al. 2017) and effects on carbon sequestration (Nguyen et al. 2010) amongst others remain areas for debate. A summary of the challenges in characterising the impacts from land use and land use change are given by Nordborg et al. (2017). However, the authors of this paper regard LUC as more of an end-point issue and for that reason, as well as the complexities involved which we regard as out of scope, this article will focus only on land use rather than the effects of change.
2 Goal and scope
The aim of this paper is to view and compare the contributions to the environmental impacts of Scottish salmon production through process-based and geographical-based lenses to offer insights and contextualisation. Particularly, we identify emissions and demands on resources, which may be of more concern geographically rather than on a global basis, and compare local to global impacts. This may aid in identifying the most pertinent and achievable areas for improvement than contributions from processes alone.
2.2 Functional unit, boundaries, allocation and uncertainty
The functional unit of this study is 1.0 t of head-on gutted (HOG) Atlantic salmon, boxed in ice at the primary processor gate. Primary data collection included the feed mill, production sites and the primary processor, whereas feed ingredient production and processing data were from literature sources. More details on the data collected are given in Sect. 3.
According to ISO 14040 and 14044 guidelines (ISO 2006a and 14044b), this study subdivided multi-functional processes where possible, but used economic allocation for most multi-functional processes throughout, as there is a mix of near-waste by-product production (e.g. viscera at the primary processor) and co-product production with several “target products” in other cases (such as fishmeal and fish oil at the reduction plant). Economic allocation was used as being most appropriate given the motivations of the industry at each point in the value chain and the dynamics of transitioning between waste and co-production (MacKenzie et al. 2017). However, as co-product allocation is often a contentious subject and results may differ considerably according to the method chosen (Pelletier and Tyedmers 2007; Ayer et al. 2007), a sensitivity analysis, using mass allocation, has been included in the supporting information.
This study presents contribution analyses only; therefore, no uncertainty is presented on impact values for any category. However, a Numerical Unit Spread Assessment Pedigree (NUSAP) was used to account for representativeness during horizontal averaging of literature data according to Henriksson et al. (2014a).
2.3 Assessment method and impact categories
This study uses a CML2001 mid-point attributional approach. All modelling was conducted on CMLCA software licenced by Leiden University using the EcoInvent 2.2 database for background data such as transport emissions. Impact categories used were chosen according to the most important categories for aquaculture according to Pelletier et al. (2007) which were as follows: global warming potential (GWP), acidification potential (AP), eutrophication potential (EP), ozone depletion potential (ODP), photochemical oxidation (PCO), with the addition of consumptive water use (CWU) and land use (LU). These impact categories are most suitable for measuring the cumulative impact of aquaculture and the trade-offs within the value chain from a local and global perspective. However, toxicity potentials were not included because many of the chemicals and therapeutants used in aquaculture have not been well characterised for LCA purposes. Similarly, net primary productivity (NPP) was not included because methods are still not well defined (Cashion et al. 2016), and often, the category does not indicate the sensitivity of biotic resources. Impacts on biodiversity are not included because in many circumstances, the effects are difficult to measure and/or characterise throughout the VC (e.g. Ford et al. 2012). A distinction was made between CWU and water dependence, where CWU was regarded as water that was consumed by the system and no longer available for other uses, irrespective of quality changes. Water dependence is a different issue and refers to water that an aquaculture system requires to maintain environmental services such as oxygen provision or effluent dilution. Such water may then be available for other uses such as municipal use, irrigation or industry depending on its qualities. In this case, only blue and green water use were considered, where green water is that precipitated or present in the soil and blue water is derived from irrigation. Total water requirements were calculated from crop rainfall requirements in millimetres, given by Brouwer and Heibloum (1986) and assumed to be provided either by precipitation or irrigation for this study. Rainfall was then extrapolated to the given yields per hectare to provide a CWU in cubic metres per tonne of production. The calculation was validated by applying it to available water requirement data for crops given by Mekonnen and Hoekstra (2011) which were broadly in agreement, although their figures were towards the upper end of our calculated range. Unlike Mekonnen and Hoekstra (2011), we did not consider “grey” water requirements.
3 Data collection and life cycle inventory
Scottish salmon production is based exclusively in net pens in marine sites off the West Coast mainland, and the islands of the Outer and Inner Hebrides, Orkney and Shetland. At the time of data collection (2012 to 2014), smolts (juveniles) were produced to 70 g average, in freshwater sites before transportation to sea by road and boat. The fish were kept in 90-m circumference net pens for a period of around 22 months, at which point they reached an average size of around 4.5 kg before slaughter. Fish were transported by well boat from the production site to a slaughter station which may be combined with primary processing (gutting and bleeding). At this point, the head-on gutted (HOG) salmon were sent for further processing, either domestically or abroad.
Ingredient list and origin of ingredients included in grow-out salmon feed used in this study
Country of origin
Oceanic freight/road lorry
Oceanic freight/road lorry
Oceanic freight/road lorry
Oceanic Freight/road lorry
Sunflower seed cake5,6,7
Oceanic freight/road lorry
Oceanic freight/road lorry
Vitamin and minerals4
Primary data were collected from a major Scottish-based aquafeed mill which also has representation in other major salmon-producing countries. This included individual ingredients and quantities for standard grow out feed for 1 year of supply (2012), all electricity, gas, water consumption and land area. The data provided the quantity of each ingredient in the formulation and the location from which it was sourced, although specificity of location was variable. The source of some ingredients could be located to the county level, but for others only to the broad geographical region (e.g. “continental Europe”). The individual feed ingredients and their origins have been included in Table 1, but quantities have been shown aggregated, due to the wishes of the company, into broader categories such as marine proteins, vegetable oils etc.
Feed ingredient production and processing, data were taken from literature sources which provided information on fertiliser application, fuel use in machinery, other energy use, yield per hectare and chemicals used for separation and refining (Table 1). Road and ocean transport distances were calculated from Google Maps. Of the crop production data, only a few sources included freshwater requirements for those crops. Data on fishmeal and fish oil from Peru were from Henriksson et al. (2014b) adapted from Avadí et al. (2014) whereas the Danish fishmeal and fish oil were adapted from the same process with the energy mix and transport adjusted accordingly.
Farm data was collected from six individual sites, although some shared land bases and some other resources, such as boat use. Data were collected on quantity of feed used, electricity use, diesel for boats and service barges, fish mortality, medications, road distances from the shore base to the feed supplier, land occupation of the shore base and total salmon production. The weighted mean economic feed conversion ratio (eFCR) between the six farms was calculated at 1.19 ± 0.10 kg of feed per kg of fish growth including fish losses. Medications for salmon production have not been widely characterised, and in this case, a pesticide process was used from the Ecoinvent database (Frischknecht et al. 2005). Data on the production of smolts were not collected in this instance; it was assumed that their contribution to overall impacts would be minimal, being only 1.5% of the size of the final product at stocking. Finally, data was collected on the primary processing of salmon, which included gutting and packing in ice within polystyrene boxes, prior to transport for further processing.
Feed is by far the most important contributor to impacts throughout the value chain (Fig. 1), as the embodied impacts of feed production account for more than 80% of all impact categories, and this finding is supported in other salmon LCAs (e.g. Ayer and Tyedmers 2009; Pelletier et al. 2009; McGrath et al. 2015). An exception is EP, where the majority of impacts come from the metabolites of salmon growth, on the production unit. Apart from feed, notable impacts on ozone depletion potential occur from chemical usage at both the farm and at the processor. On the farm, this is connected to the production of medicinal treatments, whereas at the processor, it is the use of refrigerants which is of most concern. Minor but significant contributions to ozone depletion potential also occur through diesel used on the service barges and boats at the farm site. Notable contributions also occur from packaging materials at the processor, to PCO and GWP. However, these contributions are dwarfed by the direct and indirect contributions from feed production through to the salmon raw material at the processor. Contributions are often affected by the allocation procedure that is adopted (e.g. Pelletier and Tyedmers 2007), and so, a sensitivity analysis was conducted using mass allocation to test whether this had a significant influence on either the overall impact or the contributions to those impacts, particularly from different feed ingredients. The detailed results of this analysis can be found in the supporting information (Figs. S1 and S2, Tables S1 and S2, Electronic Supplementary Material), but generally, there were minor differences, both in the overall impact and in the contributions between the two allocation methods. Generally, mass allocation showed slightly lower overall impacts in all categories except land and consumptive water use which were higher using mass allocation. Consequently, the contributions to these two categories change most, from both the input processes and geographically, between the two allocation methods.
In Fig. 2, the geographic spread of emissions highlights disproportionate impacts from PCO in Brazil and from AP in continental Europe. The PCO contributions can be matched to vegetable protein production in Brazil which is attributable to land management and refinement of soy protein. High AP contributions in Europe are mostly linked to SO2 emissions from high energy inputs required for the processing of wheat and corn gluten, even though it makes up a relatively small proportion of the feed formulation. Whereas GWP and ODP may be considered of global importance, AD and PCO are of more localised and regional significance. Europe especially has a history of problems from acid rain dating back to early industrialisation, since when emissions from Western and central European countries have been linked to acidification effects in Central and Eastern Europe, Scandinavia and further afield (Menz and Seip 2004). Although significant steps have been made to reduce SO2 emissions in Europe, disproportionate contributions such as from gluten, which contribute 36.8% to AP at the feed mill stage, may be considered to be of concern. The effects of PCO from soybean cultivation and refinement are less clear, as the emissions are likely to be in more rural areas and dispersed more easily than in cities, and where it is less of a human health concern.
While this LCA study serves to highlight geographically disproportionate emissions and demands on resources, it is limited in detail to gauge the consequence of these impacts. The significance of emissions may be on a highly localised level, e.g. sensitive water body; regional level, e.g. land mass; or global level. For localised emissions, more contextual information may be necessary on the type of emission and the local conditions to assess whether the system can absorb those impacts without having an adverse effect on public health or wildlife. An example is if sea currents are adequate to disperse both dissolved and solid waste emitted from net pen sites or at a city level, how much PCO emissions are likely to cause public health issues. In order to do this, it would be necessary to normalise data to whatever level was necessary for the context of the emission and model its dispersion within that environmental system. These were considered end-point assessments, and while beyond the scope of this article, geographic information systems (GIS) could aid in identifying such issues. There have been a few efforts to integrate LCA and GIS in such a way that adds context to the ability of environmental systems to absorb localised/regionalised impacts or emissions, notably by Gasol et al. (2011), Dresen and Jandewerth (2012) and Mutel et al. (2012), but while offering interesting findings and setting out methodology and principles; so far, they have not been widely adopted.
In terms of overall improvements, rather than substituting small amounts of feed ingredients within diets, more substantial impact reduction can be made improving feed efficiency overall. Although there may be seasonal variability due to different challenges, improved feed efficiency has been the trend in most fin-fish aquaculture as demonstrated by Tacon and Metian (2015) and more improvements are to be expected, aided by such innovations in advanced ingredient development mentioned above. Further improvements have and could be made by improved automated feed delivery technology, genetic improvements, maturation control, better stock management and reduced mortality. Mortality is included when calculating eFCR, and improved survival can therefore substantially improve environmental performance, depending on when the majority of mortality occurs. Weighted mean mortality for the six farm sites in this study was 4.0% by weight, which is below the industry average estimated at between 5 and 10% (SEPA 2004, Statistics Norway 2009) although the lowest mortality rate was just 1.0%. Mortality continues to be a challenge for the industry particularly from sea lice which not only cause losses at the farm, but also at the processor, as consignments are rejected because of poor quality if lice counts exceed a specified number. As this problem is tackled, it is likely that the dynamics and trade-offs between different impacts may change. The lowest eFCR out of the six sites was 1.04 which is a 14.4% improvement on the average, although this site did not have the lowest mortality, suggesting that further improvements could be made. This would translate into substantial environmental impact abatement, if all farms were to achieve this benchmark of feed efficiency.
Also of note, is the minor contribution of transportation to any of the impact categories apart from ODP perhaps, to which ocean transport contributes 8.1%. Such issues raise further questions as to whether so-called local production is “better” than food imports. When less than 25% of the raw materials are considered local, can the final product be considered as “locally produced”? Therefore, “food miles” should not be of concern to stakeholders from an environmental perspective, more the efficiency and responsibility of production within those countries prior to transport. That is, from an environmental impact perspective, it is preferable to source materials produced efficiently and cleanly from further afield than local materials produced inefficiently and with polluting practices. For example, Boissy et al. (2011) demonstrated that palm oil from Malaysia was preferable to German rapeseed oil in many impact categories. Although there may be genuine concern over the environmental impacts of global value chains in some circumstances, these may be unfounded and must be looked at in more detail on an individual basis.
A conscientious consumer may wish to buy locally produced food on the basis that they know it is responsibly produced from a food safety, environmental and social impact perspective. EU food law (EC 2002) and traceability generally ensure the safety of food produced within and imported to the EU, but other responsibility issues are generally only covered by third party certification such as the Aquaculture Stewardship Council (ASC 2017). However, such certification schemes tend to focus on the production unit, with limited attention to whole value chain (Bush et al. 2013). This has led to issues elsewhere in the value chain being under-investigated. A recent example was the supply of fishmeal in Thai shrimp value chains, which was criticised for poor labour practices (https://www.theguardian.com/global-development/2014/jun/10/supermarket-prawns-thailand-produced-slave-labour, accessed 10/7/17). Whereas producers engaged with certification schemes that monitor shrimp farms, at the time, the monitoring did not extend to social practices throughout the VC and failed to pick up the problem. However, aquaculture has also been credited with improving the livelihoods of many vulnerable and/or remote communities, including rural populations on the West Coast of Scotland which has beneficial knock on effects throughout the value chain and associated communities (SSPO 2015). Such benefits to the local economy may out-weigh other considerations, especially when the global impacts of imported alternatives may be regarded as similar or more difficult to gauge. The choices made regarding food production, supply of materials, the knock-on environmental and social impacts and trade-offs are highly nuanced. However, retailers who advertise their produce as “local” have a duty to ensure and communicate the responsible credentials of their product from a value chain perspective, if consumers are purchasing the product on the basis that local production is ethically superior.
From a broader policy perspective, this study identifies the inadequacies of assessing the sustainability of food systems from highly localised, national standpoints alone. The study also highlights the need for balance between use of marine and terrestrial ingredients, and their resulting environmental impacts, in driving feed formulation for commodity food products such as farmed Atlantic salmon. The implications for global trade in food ingredients on policies supporting environmental restoration and conservation in both producer and consumer countries are also important, as has been noted for the telecoupled Brazil-China soybean trade (Torres et al. 2017)
This study shows that life cycle impact assessments can be strengthened by including more geographically specific data which enable contextualisation. To produce a fully integrated LCA/GIS study, a large dataset is required that details the characteristics of the environment where the emissions occur at each point in the value chain, so that the dispersion and assimilative capacity may be calculated. Such a dataset may not be practical to collect in the short term. However, efforts are being made by Global Feed LCA Institute (GFLI), constituted by a consortium of European feed associations, to build a feed ingredient LCA database using a standardised methodology developed by the FAO LEAP project (FAO 2016). Such a dataset may provide the necessary detail to facilitate GIS integration consistently. However, in this article, we have presented a practicable geographically contextualised result set which can be used to quickly highlight impacts of geographical concern which may then be investigated in more detail.
The study shows that Scottish salmon is not truly a “local” product but supports a vast global value chain. Consumers may be unaware that their support for local production, for environmental or other responsible production motivations such as socio-economic, food safety or animal welfare, is in fact supporting production on a much broader basis, and contributing to the associated impact that this may have or indeed benefits to those regions.
- ASC (2017) ASC Salmon Standard Version 1.1 April 2017. Aquaculture stewardship council, Utrecht 103 ppGoogle Scholar
- Boulay AM, Motoshita M, Pfister S, Bulle C, Munoz I, Franceschini H, Margini M (2015) Analysis of water use impact assessment methods (part a): evaluation of water modeling choices based on a quantitative comparison of scarcity and human health indicators. Int J Life Cycle Assess 20(1):139–160CrossRefGoogle Scholar
- Brouwer and Heibloum (1986) Irrigation Water Management: Irrigation Water Needs – Part 1; Principles of Irrigation Water Needs. Chapter 2 – Crop Water Needs. Food and Agriculture Organisation of the United Nations, Rome. Available from: http://www.fao.org/docrep/s2022e/s2022e02.htm#chapter2:cropwaterneeds
- EC (2002) Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January2002 Laying Down the General Principles and Requirements of Food Law, Establishing the European Food Safety Authority and Laying Down Procedures in Matters of Food Safety. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2002:031:0001:0024:EN:PDF. 24 pp
- FAO (2006) Livestock’s Long Shadow. Food and Agriculture Organization of the United Nations (FAO), Rome 390 ppGoogle Scholar
- FAO (2016) Environmental performance of animal feeds supply chains: guidelines for assessment. Livestock environmental assessment and performance partnership. FAO, Rome 141 ppGoogle Scholar
- Henriksson P, Zhang W, Nahid S, Newton R, Phan L, Hai D, Zhang Z, Andong R, Jaithang J, Chiamanuskul K, Son V, Hien H, Haque M, Kruijssen F, Satapornvanit K, Nguyen P, Phan L T, Liu Q, Liu L, Wahab M, Murray F, Little DC, Guinée J (2014b) Final LCA case study report primary data and literature sources adopted in the SEAT LCA studies, SEAT deliverable ref: D 3.5. EU FP7 funded project 222889. Available at: http://media.leidenuniv.nl/legacy/d35-annexreport.pdf. 121 pp
- IEA (2014) International Energy Authority Statistics https://www.iea.org/statistics/statisticssearch/. Accessed 24 February 2014
- IPCC (2006) N2O emissions from managed soils and CO2 emissions from lime and urea application, IPCC guidelines for National Greenhouse gas Inventories: vol 4, agriculture, forestry and other land use, Ch. 11. International panel on climate change, Geneva 54 ppGoogle Scholar
- ISO (2006a) Environmental management - life cycle assessment - principles and framework (ISO 14040:2006). ISO, Geneva 20 ppGoogle Scholar
- ISO (2006b) Environmental Management - life cycle assessment - requirements and guidelines (ISO 14044:2006). ISO, Geneva 45 ppGoogle Scholar
- Jackson A, Newton R (2016) Project to model the use of fisheries by-products in the production of marine ingredients with special reference to omega-3 fatty acids EPA and DHA. A report by IFFO and the University of Stirling. IFFO, London 14 ppGoogle Scholar
- Janßen D, Langen N (2016) The bunch of sustainability labels - do consumers differentiate? J Clean Prod 143:1233–1245Google Scholar
- Roberts CA, Newton RW, Bostock JC, Prescott SG, Honey DJ, Telfer TC, Walmsley SF, Little DC, Hull SC (2015) A risk benefit analysis of Mariculture as a means to reduce the impacts of terrestrial production of food and energy. A study commissioned by the Scottish aquaculture research forum (SARF). http://www.sarf.org.uk/212pp
- Robinson TP, Thornton PK, Franceschini G, Kruska RL, Chiozza F, Notenbaert A, Cecchi G, Herrero M, Epprecht M, Fritz S, You L, Conchedda G, See L (2011) Global livestock production systems. Food and Agriculture Organization of the United Nations (FAO), Rome and International Livestock Research Institute (ILRI), 152 ppGoogle Scholar
- Schmidt JH (2010) Comparative life cycle assessment of rapeseed oil and palm oil. Int J Life Cycle Assess 15:183–197. SpringerGoogle Scholar
- SEPA (2004) Evaluation of fish waste manage techniques. SEPA contract reference 230/4198. Scottish Environmental Protection Agency, Edinburgh 48ppGoogle Scholar
- SSPO (2015) Sustainable Scottish Salmon Scottish Salmon farming economic report June 2015. Scottish Salmon Producers’ Organisation, Perth, 16 ppGoogle Scholar
- Statistics Norway (2009) Fishery statistics 2007, official statistics of Norway D 428. Statistics Norway, Oslo, 13 ppGoogle Scholar
- Troell M, Metian M, Beveridge M, Verdegem M, Deutsch L (2014a) Comment on ‘water footprint of marine protein consumption - aquaculture’s link to agriculture’ environ res Lett 9:1-4Google Scholar
- UNEP (2000) Cleaner production assessment in fish processing. Report prepared by COWI consulting engineers and planners AS, Denmark on behalf of United Nations Environment Programme Division of Technology, Industry and Economics and Danish Environmental Protection Agency, 99 ppGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.