Life cycle assessment of diets for gilthead seabream (Sparus aurata) with different protein/carbohydrate ratios and fishmeal or plant feedstuffs as main protein sources

The International Journal of Life Cycle Assessment https://doi.org/10.1007/s11367-019-01625-7 OCEAN RESOURCES & MARINE CONSERVATION Life cycle assessment of diets for gilthead seabream (Sparus aurata) with different protein/carbohydrate ratios and fishmeal or plant feedstuffs as main protein sources Catarina Basto-Silva 1,2 & Inês Guerreiro 1 & Aires Oliva-Teles 1,2 & Belmira Neto 3 Received: 31 January 2019 / Accepted: 3 April 2019 # Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract Purpose Aquaculture is the best alternative to fulfil global fish demand, however it still relies heavily on fisheries-derived products for aquafeeds production. This study assesses and compares the environmental impacts of producing four experimental diets to gilthead seabream with different dietary protein (P) to carbohydrate (CH) ratios (P50/CH10 and P40/CH20). The diets were made either with fish meal (FM) or plant feedstuffs (PF) as main protein sources and fish oil (FO) or vegetable oils (VO) as lipid sources. Methods The functional unit used was 1 kg of experimental diet. The studied boundaries included aquafeed ingredients production (S1), compound aquafeeds production under laboratory conditions (S2), and transportation between S1 and S2 locations. The present study applied the Recipe Endpoint method, hierarchist version (V1.13; Europe recipe H/A). Background data was collected from ecoinvent database and related literature. For each aquafeed ingredient used, it was accounted either the agriculture production or fishery activities, the processing unit, and transportation between the production and processing locations. Ingredient mixing and processing was done at the Marine Zoology Station (MZS) located at Porto, Portugal. It was also taken into account the road transportation of aquafeed ingredients between a commercial company and the MZS. Results and discussion Regardless of dietary protein source or P/CH ratio used, all diets had the same single score index. In agreement with several studies, S1 was the system with the highest environmental impact. On the other hand, S2 was the lowest environmental contributor step to all formulated diets, except for diet P50/CH10, where the lowest environmental impact was related with the aquafeed ingredients transportation to MZS. Fisheries-derived ingredients were the biggest contributors to environmental impact. In the hypothesis of replacing FO from Portuguese fisheries by-products by FO of Peruvian anchovy fisheries or by soybean oil (SBO), the environmental impact of the diets would be decreased, being the replacement by SBO the best environmental alternative. Responsible editor: Ian Vázquez-Rowe Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11367-019-01625-7) contains supplementary material, which is available to authorized users. * Belmira Neto belmira.neto@fe.up.pt 1 CIIMAR - Centro Interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto, Terminal de Cruzeiros do Porto de Leixões, Av. General Norton de Matos s/n, 4450-208 Matosinhos, Portugal 2 Departamento de Biologia, Faculdade de Ciências, Universidade do Porto, Rua do Campo Alegre s/n, Ed. FC4, 4169-007 Porto, Portugal 3 LEPABE – Laboratório para Engenharia de Processos, Ambiente, Biotecnologia e Energia, Faculdade de Engenharia, Universidade do Porto, R. Dr. Roberto Frias s/n, 4200-465 Porto, Portugal Int J Life Cycle Assess Conclusions There was highlighted a tendency for PF-based diets having lower environmental impact score when compared to the FM-based diets after both hypothetical replacements. Studying the replacement of by-products FO by Peruvian anchovy FO or by SBO allowed to emphasize the importance of adequate ingredients selection for reducing the environmental impact. As limitations of the current study it is important to mention: the use of pelletization as manufacturing process and the nonvalorization of fish by-products as recycled aquafeed ingredients. Keywords Aquafeed ingredients . Environmental impacts . Experimental diets . Plant feedstuffs . Recipe endpoint method . Sustainable aquaculture 1 Introduction It is estimated that by 2050 nine billion people in the world will need to be fed (FAO 2018). Fish is a good candidate to fulfil animal protein needs since it is easily digested; rich in essential amino acids; rich in several vitamins and minerals such as vitamin D and A, calcium, iodine, zinc, iron, and selenium; and rich in long-chain omega-3 fatty acids. In addition, a healthy diet might also prevent some diseases of the 21st century, such as obesity, cardiovascular diseases, and malformations of nervous system during foetal and infant development (FAO 2016; FAO 2018). In 2015, fish represented about 17% of all animal protein consumed by the world population, being the third major protein source in human diet, after cereals and milk, and represented also 7% of all protein consumed (FAO 2018; Tacon and Metian 2018). Tacon and Metian (2018) stated that due to the great demand for fish protein, aquaculture will be the only real solution to supply global market needs, as capture fisheries stabilized since the 1990s. In fact, aquaculture had an annual average growth of 8% over the past 30 years, having the biggest annual growth compared to the other food commodities (Naylor 2016; FAO 2018; Tacon and Metian 2018). In 2014, for the first time, aquaculture production for human consumption surpassed the amount of fish provided by captures (FAO 2016). However, modern aquaculture has two main sustainability problems, one related with land occupation and other with compound feeds (Edwards 2015). Due to its nutritional composition, fishmeal (FM) and fish oil (FO) are considered the most adequate protein and lipid sources, respectively, to be used in aquafeeds for carnivorous fish (Tacon and Metian 2008; 2015). In addition, these ingredients are highly digestible and have good palatability (Rust et al. 2011; Olsen and Hasan 2012). In 2015, approximately 70% of FM and 73% of FO world production was used in aquafeeds. Marine fish aquaculture is the third larger user of FM, using 18% of overall production, just after shrimp and salmonids with, respectively, 30% and 22%. Regarding FO, marine fish are the second larger users, just after salmonids, with 23% and 58% consumption, respectively (IFFO 2016). Nonetheless, FM and FO inclusion in aquafeeds decreased over the last years, due to reduction and/or stabilization of wild fisheries stocks available for FM and FO production; increase of small pelagic fish prices, due to increased fishing costs and direct human consumption; increase of FM and FO prices in the global market; and increased market and social pressure on feed manufacturers to replace FM and FO in aquafeeds by more sustainable alternatives (Tacon and Metian 2008; Olsen and Hasan 2012). The limited amount of available FM and FO for aquafeeds lead to an increased research effort to find alternative protein and lipid sources (Olsen and Hasan 2012). A reduction of fisheries ingredients usage in aquafeeds may be achieved by increasing the dietary incorporation of plant feedstuffs (PF), which appear the best alternatives to FM and FO at the moment, or by partially replacing dietary protein (P) content by carbohydrates (CH) (Robaina et al. 1995; Enes et al. 2008; Montero et al. 2008; Dias et al. 2009; NRC 2011; Izquierdo et al. 2015). However, besides reducing the impact on fish wild stocks, the environmental benefits of replacing FM/FO by PF sources are not clear. According to Silva et al. (2018) soybean meal (SBM) and soybean oil (SBO) and FM/FO from Peruvian anchovy seem to have similar environmental impacts. Moreover, more studies will be needed to understand the environmental impact of partially replacing dietary protein by CH. Life cycle assessment (LCA) allows a comparison and evaluation of products’ environmental impacts throughout their life cycle. This methodology is described in ISO 14040 (2006) and includes three mandatory steps: (i) goal and scope definition (which allows product and system description, and their boundaries definition), (ii) life cycle inventory analysis (LCI) (which embraces data collection and validation, as well as calculation of environmental load), (iii) life cycle impact assessment (LCIA) (which includes classification, characterization, normalization, and weighting previous collected data), and (iv) interpretation step (where are presented the main results, conclusions, and/or recommendations). LCA approach assumes an important role to ensure aquaculture sustainability, since aquaculture production increases continuously, representing already, in 2016 about 47% of all fish produced through fisheries and aquaculture (FAO 2018). Several studies focusing on aquaculture LCAs have already been published (Iribarren et al. 2012; Aubin 2013; García et al. 2016; Abdou et al. 2017; Avadí et al. 2018). Some concluded that feed production is the main contributor to the environmental impact when compared to the remaining phases of fish Int J Life Cycle Assess production life cycle. Nevertheless, only a few studies focused on assessing the environmental impacts associated with aquafeeds formulation and production (Boissy et al. 2011; Samuel-Fitwi et al. 2013; Avadí et al. 2015; Strazza et al. 2015; García et al. 2016; Silva et al. 2018). For instance, Boissy et al. (2011) compared Atlantic salmon (Salmo salar) and rainbow trout (Oncorhynchus mykiss) standard diets with a low fishery product diet and concluded that in both cases low fishery product diet had relatively higher environmental impact for most categories. On the contrary, Samuel-Fitwi et al. (2013) concluded that dietary replacement of FM by soybean and rapeseed meals significantly reduced the environmental impact of producing one tonne of rainbow trout aquafeed. Avadí et al. (2015) compared the environmental impact of artisanal (simple technology, small scale, simple formulations, using local inputs) and commercial technology to produce one tonne of rainbow trout, tilapia (Oreochromis spp.), or black pacu (Colossoma macropomum), and concluded that commercial aquafeeds had the highest environmental impact. Strazza et al. (2015) evaluated the environmental impacts associated with the production of conventional feeds for salmon in UK and Norway, concluding that the worst carbon footprint and non-renewable energy demand was found in UK, while the highest impact for water scarcity was found in Norway. García et al. (2016) focused on gilthead seabream (Sparus aurata) production, and concluded that FM and SBM were the ingredients with the highest contribution to the measured impact categories. On the contrary, Silva et al. (2018) concluded that from an LCA perspective, SBM/SBO and FM/FO from Peruvian anchovy seem to be interesting options for aquafeeds. However, to the author’s knowledge, no study evaluated the environmental impacts associated to diets with different P/CH ratios. In Europe, gilthead seabream is one of the most economically important aquaculture species. In 2016, it was the ninth major species produced in Europe, representing 2.4% of aquaculture production (FAO 2019). In 2016, about 186 million tonnes of gilthead seabream were produced worldwide, with around 58% being produced in Turkey and Greece, whereas Portugal produced only about 1162 tonnes, which represent less than 1% of world overall production (FAO 2019). The aim of the present work was to evaluate and compare the environmental impacts associated to producing laboratory compound aquafeeds for gilthead seabream with different dietary protein sources (FM or PF) and different P/CH ratios. This was done to cover the current scientific gap on the environmental impacts of gilthead seabream diets, which may be also extrapolated to other aquaculture fish species fed with similar diets. In addition, it will allow to analyze potential environmental improvements by the use of alternative ingredients. 2 Materials and methods 2.1 Goal and scope This work aims to identify diets with lower environmental impact to support decision making. Thus, this study evaluated and compared the environmental impacts of producing four experimental diets for gilthead seabream. The diets were formulated with two different protein sources and two P/CH ratios and were made at laboratory scale, for investigation purposes. LCA approach followed the concept Bfrom cradle to gate^ and considered two systems. The first system relates to the production of aquafeed ingredients (S1) and the second to the production of the diets (S2), which includes mixing and pelletization of feed ingredients. All ingredients are commercialized by an industrial company located at Ovar (Portugal), except pregelatinized maize starch and cellulose, which are commercialized by a company from Mechelen (Belgium) and Sintra (Portugal), respectively. Transportation of the aquafeed ingredients to the Experimental Marine Zoology Station (MZS) facilities at Porto (Portugal), where the diets were made, was considered in this study. Based on preliminary assays using the laboratory oven, and due to the low energetic efficiency of the equipment, the pellet drying (within S2 stage) was excluded from this LCA. It was also excluded from the analysis the construction of infrastructures and equipment, end-of-life of capital goods, and wastes from administration, laboratory, canteen, or offices. This exclusion was made since this study was done at laboratory scale and therefore it would strongly bias data evaluation, also because most of the studies exclude these items from analysis, allowing us a better comparison and discussion of the results. The functional unit (FU) is 1 kg of experimental diet produced. According to the ISO 14044 (2006) some hierarchy steps were performed to address allocation. The LCA approach is attributional and we used mass allocation due to the following reasons: most studies available in the literature use mass allocation, thus allowing an easier comparison between studies (Iribarren et al. 2012; Samuel-Fitwi et al. 2013; García et al. 2016); high prices fluctuations for ingredients does not allow a steady economic allocation, mainly for FM and FO used in diets (Tacon and Metian 2008); and energy allocation is not useful, since aquaculture farmers often use mass unit to evaluate production performance, as for example, the amount of feed intake per fish or tank, growth efficiency, or feed conversion ratio (Martínez-Llorens et al. 2007). A more detailed description of system boundaries is presented in Fig. 1. 2.2 Diets description This study assessed four isolipidic (18% crude lipids) experimental diets that were formulated at the MZS using commercial feed ingredients. The diets had two P/CH ratios and Int J Life Cycle Assess Fig. 1 Boundary system considered to each of the formulated diets. Functional unit: 1 kg of experimental diet (MZS: Experimental Marine Zoology Station) Infrastructure AQUAFEED INGREDIENTS PRODUCTION (S1) Equipment Transport of aquafeeds until MZS Fishing activities or Agriculture production End-of-life of capital goods PROCESSING UNIT (S2) Drying 1 KG OF DIET Mixing Pelletization Transport Processing unit System boundary Label: Excluded from this study different protein sources. Diets were either FM-based or PFbased (20% FM + 80% PF). PF used as protein sources were SBM, maize, and wheat glutens. The following P/CH ratios were tested: P40/CH20 or P50/CH10. Diets were formulated to fulfil gilthead seabream nutritional requirements, thus having similar nutritional composition. The ingredients and proximate composition of the experimental diets are presented in Tables 1 and 2, respectively. The nutritional composition of feed ingredient used is available on footnotes of Table 1. The origin of each ingredient was made available by each company. The proximate composition was determined for (i) aquafeed ingredients and (ii) experimental diets, following methods from the Association of Official Analytical Chemists (AOAC 2000). FM, SBM, maize gluten, and wheat gluten have different protein and lipid contents. However, FO is 100% lipids, thus this ingredient does not need to be analyzed. Ingredients and diets were dried in an oven at 100°C until constant weight for moisture content determination. Ash matter was determined by incineration in a muffle furnace at 450°C, during 16 hours. Protein content followed the Kjeldahl method, using a Kjeltec digestor and distillation units (Tecator Systems, Höganäs, Sweden; model 1015 and 1026, respectively). Lipid content was analyzed by petroleum ether extraction using a SoxTec extraction system (Tecator Systems; extraction unit model 1043 and service unit model 1046). Dietary starch content was determined according to Beutler (1984) and rendered the amount of energy storage carbohydrates present on the formulated diets. 2.3 Aquafeed ingredients (S1) Ingredient selection was made according to the nutritional values of each ingredient, availability, and economic costs. Included in this study System boundary considered FM was originated from Chile anchovy fisheries. Due to the lack of information, and geographic nearness between Chile and Peru, this study uses the information collected by Fréon et al. (2014, 2017), which focus on anchovy fisheries and FM production in Peru. According to Fréon et al. (2017) there are three different qualities of FM in Peru: (i) Super-Prime (having more than 67% of crude protein (CP)), (ii) Prime (with 65– 67% of CP), and (iii) Standard or residual FM (produced from fish residues, often with less than 65% of CP). According to this classification, FM used in the present study was from super-prime quality (with about 77% CP). Fréon et al. (2017) assumed that super-prime FM and prime FM production had the same environmental impact. In addition, between fishing dock and FM processing unit in Peru there is no road transportation, since the fresh fish goes into the processing unit through a pumping system (Fréon et al. 2017). It was considered the maritime transport between Peru and the Netherlands, and the road transport from the Netherlands to Portugal (Ovar), as described by Silva et al. (2018). FO was derived from fisheries by-products produced in Portugal. As described in Silva et al. (2018), two stages for FO production were considered: the fishing activities and the by-products processing. The main species used for FO production were sardine and tuna, being usually captured by purse seine fishing (as described by Silva et al. (2018)). The canning industry, closely located to the fishing dock, receives the fresh fish and removes heads, tails, and guts before fitting the fish into canning boxes (Almeida et al. 2015). According to Silva et al. (2018), fish by-products from canning industry went to FM/FO production without any industrial processing. Thus, in the present study, it was considered the transport of fish between the fishing dock and the canning industry, but it Int J Life Cycle Assess Table 1 Ingredients used (% dry weight) to formulate the experimental diets Diet Protein source 100% fishmeal (FM) 80% plant feedstuffs (PF) 20% fishmeal (FM) P/CH ratio P40/CH20 FM P40/CH20 P50/CH10 FM P50/CH10 P40/CH20 PF P40/CH20 P50/CH10 PF P50/CH10 Chile 51.9 64.8 10.4 13.0 Cellulosee Monocalcium phosphate Portugal Brazil Portugal Brazil Belgium Portugal Portugal 11.9 – – – 20.0 12.7 – 10.4 – – – 10.0 11.3 – 15.7 19.1 20.0 9.0 16.6 2.9 2.1 15.2 25.0 22.6 12.7 5.9 – 1.5 Lysine Taurine Vitamin mix Portugal Portugal Portugal – – 1.0 – – 1.0 0.5 0.2 1.0 0.6 0.2 1.0 Mineral mix Binder Choline chloride (50%) Portugal Portugal 1.0 1.0 0.5 1.0 1.0 0.5 1.0 1.0 0.5 1.0 1.0 0.5 Ingredients (% dry weight) Fishmeala Fish oil Soybean mealb Maize glutenc Wheat glutend Pregelatinized maize starch Origin CP crude protein, DM dry matter, GL gross lipid a CP 77.1% DM, GL 11.8% DM b CP 52.0% DM, GL 1.9% DM c CP 70.1% DM, GL 2.8% DM d CP 83.1% DM, GL 1.4% DM e CAS no. 9004-34-6 was not accounted the environmental impact of the canning industry activity. The by-products from canning industry are collected and transported to the processing unit at Ovar to produce FO (Silva et al. 2018). Table 2 Results from laboratory essays for the approximate nutritional composition of experimental diets (% dry weight basis) According to Cavalett and Ortega (2009), soybean production in Brazil can be classified in two models: Southern lands (domestic and smaller system) and Cerrado (monoculture system production, using huge amounts of machinery and Diets Protein source 100% fishmeal (FM) 80% plant feedstuffs (PF) 20% fishmeal (FM) P/CH ratio (% dry weight) Dry matter Crude protein Crude fat Starch Ash P40/CH20 FM P40/CH20 P50/CH10 FM P50/CH10 P40/CH20 PF P40/CH20 P50/CH10 PF P50/CH10 92.9 39.1 18.6 17.2 7.5 92.1 51.3 18.7 9.0 8.6 90.3 38.0 18.4 18.2 5.6 93.8 50.6 18.7 11.4 6.4 Int J Life Cycle Assess economic resources). In 2004, 75% of soybean produced in Brazil came from Cerrado (Cavalett and Ortega 2009). Thus, for this study it was assumed that SBM used derived from Cerrado. Brazilian SBM production was described according to Cavalett (2008). The author assessed soybean agriculture production, transportation to the processing unit, and processing activity to produce a final product (SBM). The present study included train transportation between processing unit and Santos dock, in Brazil, the maritime transport to the Netherlands and road transport between the Netherlands and Ovar, as described by Silva et al. (2018). The maize gluten used in the present study came from Portugal. Thus, for this study, it was considered ecoinvent database (at global level), which included agriculture production (including seeds, fertilizers applied, pesticides, and all machine operations, as soil cultivation, transport of seeds, fertilizers, and pesticides to the field (15 km), sowing, fertilization, weed control, pest and pathogen control, plant cutting, loading, transport to farm, and discharge into silo), and processing of maize gluten. Several studies focused on wheat or wheat-based product production (Deng et al. 2013; Achten and Acker 2015; Van Stappen et al. 2015; Fantin et al. 2017; Gundoshmian et al. 2017). Nonetheless, to our knowledge no one describes the production of wheat gluten from Brazil, which was the one used in the present study. According to Achten and Acker (2015), the professional LCI databases for wheat production are more complete than most studies available in literature, since some present unclear information and others lack information. Then, the present study used the database from ecoinvent for wheat agriculture production (seeds, mineral fertilizers, pesticides, irrigation water, and all machine operations) and wheat processing unit (as crushing or milling, heat treatment, dosing, mixing, and pelletization) as an average from the global level. Due to the lack of information about pregelatinized maize starch, cellulose, and monocalcium phosphate, the ecoinvent database was also used for those ingredients. For all ingredients listed above it was taken in account the transportation between the commercial company and the MZS facilities, located at Porto. All ingredients used in the formulated diets with a contribution of less than 1% of dry weight (lysine, taurine, vitamin mix, mineral mix, binder, choline chloride) were not evaluated in the present study. 2.4 Processing unit (S2) The diets were made in a laboratory context. To produce 1 kg of each experimental diet, all dietary ingredients were thoroughly mixed in a mixer equipment, with 5% of water added, during about 2 minutes per FU. Pelletization was made in a laboratory pellet mill equipment through a 2-mm die. Pelletization takes around 9 minutes per FU. 2.5 LCI The present study used SimaPro (version 8.5.2.0) and ecoinvent database (version 3.3) as the principal source of background data. For the ecoinvent processes was used the allocation at the point of substitution (APOS). The inventory data for FM was adapted from Fréon et al. (2014) to fishing activities, from Fréon et al. (2017) for the processing unit, and from Silva et al. (2018) to the transportation step between Peru and Portugal (Ovar). Due to the fact that the functional unit mentioned in the studies from Fréon et al. (2014, 2017) relate to a single product, no allocation between co-products was done in their studies. The values sourced in the mentioned literature were adapted as already described by Silva et al. (2018). All inventory data for FO was adapted from Silva et al. (2018). The SBM inventory was retrieved from Cavallett (2008) and its transportation between Brazil to (Ovar) Portugal was adapted from Silva et al. (2018). The inventory data for the remaining aquafeed ingredients (such as wheat gluten, maize gluten, pregelatinized maize starch, cellulose, and monocalcium phosphate) were taken from ecoinvent database (version 3.3). Data related with transportation between the company that commercialized the ingredients and the MZS facilities was adapted from MapQuest (2018). To transportation of the ingredients it was assumed as worst-case scenario a freight by lorry 7.5–16 tonnes. Ingredients mixing and the pelletization stage were done at the MZS facilities. The mixing stage used a mixer (Alexanderwerkand mixer with ref. 385188-W7). The nominal electricity power is within 1.1–1.3 kW and the ingredient mix duration was 2 minutes per FU. The pellets were made in a pellet mill equipment (California Pellet Mill equipment, Model CL type 5) with a power of 0.46 kW. The time needed to produce 1 kg of experimental diet was estimated as 9 minutes. As energy source was considered the most recent electricity mix for Portugal (IEA 2016). Table 3 describes the inventory table for each diet. 2.6 LCIA It was applied the Recipe Endpoint method, hierarchist version (V1.13; Europe recipe H/A). This method is available in SimaPro (version 8.5.2.0). This method allows to obtain a total single score (expressed in points - Pt), which is calculated by weighting the three endpoint categories. As mentioned in Hofstetter (1998) this study considers the average weighting factors of 40% to human health endpoint category, 40% to ecosystem quality endpoint, and 20% to natural resources category. Thus, a larger total single score expressed a larger environmental impact. Recipe method is not used often on scientific publications on seafood issues (Carvalho et al. 2014). Int J Life Cycle Assess Table 3 Inventory data for each experimental diet by functional unit: 1 kg of experimental diet (P: protein; CH: carbohydrates; MZS: Experimental Marine Zoology Station) Diets Protein source: 100% fishmeal (FM) Inventory sources Inventory origin (or references) Peru 80% plant feedstuffs (PF) 20% fishmeal (FM) P/CH ratio Input P40/CH20 P50/CH10 P40/CH20 P50/CH10 FM P40/ CH20 FM P50/ CH10 PF P40/ CH20 PF P50/ CH10 Unit S1—aquafeed ingredients (% dry weight) Fishmeal kg 0.52 0.65 0.10 0.13 Fish oil from fisheries by-products Soybean meal Maize gluten kg 0.12 0.10 0.16 0.15 Fréon et al. (2014) and Fréon et al. (2017) Silva et al. (2018) kg kg – – – – 0.19 0.20 0.25 0.23 Cavalett (2008) Ecoinvent (version 3.3) Brazil Global (RoW/production /Alloc Def, U) Wheat gluten kg – – 0.09 0.13 Ecoinvent (version 3.3) Pregelatinized maize starch Cellulose kg 0.20 0.10 0.17 0.06 Ecoinvent (version 3.3) kg 0.13 0.11 0.03 – Ecoinvent (version 3.3) Global (RoW/production /Alloc Def, U) Global (RoW/production /Alloc Def, U) Global (RoW/production /Alloc Def, U) Monocalcium phosphate kg – – 0.02 0.02 Ecoinvent (version 3.3) Water All transportations until MZS S2—processing unit Energy use: Mixing cm3 kg km 50.00 453.22 50.00 261.37 50.00 151.26 Ecoinvent (version 3.3) Ecoinvent (version 3.3) kWh 0.04 0.04 0.04 0.04 IEA (2016) Pelletization Output Diet kWh 0.07 0.07 0.07 0.07 IEA (2016) kg 1.00 1.00 1.00 1.00 50.00 360.57 However, according to Carvalho et al. (2014), the Recipe method is one of the most flexible methods available to analyze the life cycle of some products. The single score index gives a general vision of the products, allowing a faster and easier interpretation and comparison among products. To the purpose of easy, a comparison of results for the CML-IA method with other related published literature was included in Electronic Supplementary Material. 3 Results and discussion This section is divided in two parts; the first presents and discusses the results obtained, and the second analyzes impact Portugal Global (GLO/production /Alloc Def, U) Global Freight by lorry 7.5–16 tonnes Alexanderwerk mixer Callifornia Pellet mill changes of replacing FO from fish by-products by FO from Peruvian anchovy or SBO. 3.1 Results and discussion Table 4 shows the environmental contribution by stage and the total impact obtained for each formulated diet. Regardless of dietary protein source or P/CH ratio, the environmental impact indicator had the same order of magnitude for all diets. The variation between the highest (0.540) for diet PF P40/CH20 and the lowest (0.454) for diet FM P50/CH10, environmental impact single score was only 0.086 points. Production of aquafeed ingredients (S1) was the main environmental Int J Life Cycle Assess Table 4 Environmental contribution (%) by stage to each formulated diet and the total single score, using the method Europe Recipe H/A V1.13 (P: protein; CH: carbohydrates; MZS: Experimental Marine Zoology Station) Diets Protein source 100% fishmeal (FM) 80% plant feedstuffs (PF) 20% fishmeal (FM) P/CH ratio P40/CH20 FM P40/CH20 P50/CH10 FM P50/CH10 P40/CH20 PF P40/CH20 P50/CH10 PF P50/CH10 S1—aquafeed ingredients 97.7 98.1 98.1 98.7 FM from Peruvian anchovy 15.9 21.8 2.9 3.8 Fish oil from fisheries’ by-products Soybean meal 65.0 – 62.1 – 78.9 5.8 79.9 7.9 Maize gluten Wheat gluten – – – – 0.8 3.4 0.9 5.0 Pregelatinized maize starch 5.8 3.2 4.5 1.7 Cellulose Monocalcium phosphate 13.2 – 12.9 – 2.8 1.0 – 0.7 All transportations until MZS 1.5 0.9 1.1 0.5 S2—processing unit Mixing Pelletization Total single score (Pt) 0.9 39.4 60.6 0.499 1.0 39.4 60.6 0.454 0.8 39.4 60.6 0.540 0.8 39.4 60.6 0.513 Results reported to the functional unit: 1 kg of experimental diet impact contributor for all diets. S1 contribution ranged between 97.7% and 98.7% to diets FM P40/CH20 and PF P50/CH10, respectively. On the other hand, all transportations to MZS and S2 system never contributed with more than 2% for the environmental impact. Our results are in agreement with García et al. (2016) and Abdou et al. (2017) which also focused in gilthead seabream and concluded that aquafeeds production was the major contributor to the environmental impacts. This is true, when compared with the other stages, namely diet manufacturing, aquafeed ingredients transportation, fish feed production, infrastructure and equipment, fry production, and energy used. Aquafeed ingredients production was also the major environmental contributor referred in other studies with different fish species, such as salmon, rainbow trout, turbot (Scophthalmus maximus), and seabass (Dicentrarchus labrax) (Boissy et al. 2011; Iribarren et al. 2012; Samuel-Fitwi et al. 2013; Abdou et al. 2017). FO from fisheries by-products was the ingredient presenting the largest environmental contribution to all diets. Similar relevance of FO was also reported by Samuel-Fitwi et al. (2013), where FM and FO were the main contributors to all environmental impact categories, except for the eutrophication potential, where wheat starch production was the major contributor. According to Silva et al. (2018), FO from fisheries by-products production required a large amount of material and energy. For instant, 45 tonnes of caught fish were needed to produce 1 tonne of FO from fisheries by-products. Our results showed that monocalcium phosphate, maize gluten, wheat gluten, and pregelatinized maize starch were the ingredients presenting the lowest environmental impacts. Monocalcium phosphate and maize gluten presented the lower contribution, with about 1% to both PF-based diets. These results are in agreement with those of García et al. (2016), where gluten meals (maize and wheat) also showed the lowest contribution to the diets. Deng et al. (2013) also concluded that to produce 1 tonne of wheat flour, 1.3 tonnes of wheat grains were needed. Moreover, Henriksson et al. (2017) draw attention to the fact that protein sources had higher environmental impacts than carbohydrate sources. However, more studies need to be conducted for a better understanding of the environmental impact of gluten sources from different geographic regions. 3.2 Analysis of the replacement of FO from fish by-products by FO from Peruvian anchovy or by SBO As FO from fisheries´ by-products was the aquafeed ingredient with the highest environmental contribution in all diets, we evaluated the replacement of this ingredient by FO from Peruvian anchovy or by SBO, which were shown to be more environmentally sustainable ingredients (Silva et al. 2018). FO from fisheries by-products was fully replaced by FO from Peruvian anchovy. However, in order to satisfy fish nutritional Int J Life Cycle Assess requirements, SBO was only included at 90% and the remaining 10% were replaced by FO from Peruvian anchovy. Tables 5 and 6 present the results obtained by stage for each diet after replacing FO from by-products by FO from Peruvian anchovy or by SBO, respectively. Based on the total single score, all original diets had the environmental impact reduced when FO from by-products was replaced by FO from Peruvian anchovy or by SBO. After FO replacement by FO from Peruvian anchovy, the environmental impact was decreased by 54%, 51%, 65%, and 66% respectively, for diets FM P40/ CH20, FM P50/CH10, PF P40/CH20, and PF P50/CH10. Other authors presented similar conclusions when comparing the impact of fishery by-product ingredients with ingredients obtained directly from fisheries (Pelletier and Tyedmers 2007; Silva et al. 2018). Silva et al. (2018) showed that fisheries byproduct ingredients had higher environmental impacts in all evaluated environmental categories than ingredients directly obtained from fisheries. Pelletier and Tyedmers (2007) also reached the same conclusions regarding salmon aquafeeds. In the scenario of replacing FO from fisheries byproducts by SBO, the results were improved when compared with the initial diets or even with the previous scenario. Comparing with the original diets, the environmental impact after replacement by SBO decreased by 56%, 54%, 68%, 69% for diets FM P40/CH20, FM P50/CH10, PF P40/ CH20 and PF P50/CH10, respectively. Silva et al. (2018) also concluded that SBO was one of the best options for aquafeeds when compared with FO from fisheries byTable 5 Environmental contribution (%) by stage to each experimental diet and the total single score, after replacing FO from by-products by FO from Peruvian anchovy, using Europe Recipe H/A V1.13 method (P: protein; CH: carbohydrates; MZS: Experimental Marine Zoology Station) Protein source products or poultry fat. However, we should keep in mind that this environmental impact analysis does not take into account the potential benefits of fish by-products valorization, which otherwise would be wasted, with consequent economic value loss (Silva et al. 2018). Despite the total single score of all diets being of the same order of magnitude, independently of the dietary protein source or P/CH ratio used, there was a reduction of the overall environmental impact in both FO fisheries by-products substitution scenarios. It was also observed a tendency for PF-based diets to present lower environmental impacts than FM-based diets. S1 is still the largest environmental impact contributor in all diets. However, comparing with the original diets, S2 contribution increased when using FO from Peruvian anchovy or SBO. This was due to the decreased absolute value of environmental impacts associated with S1 after replacing FO from by-products by FO from Peruvian anchovy or by SBO. After replacing FO from fisheries by-products by FO from Peruvian anchovy, fisheries-derived ingredients still presented the largest contribution to all diets’ environmental impact. Regarding the FM-based diets, FM from Peruvian anchovy became the ingredient with the relatively highest contribution to environmental impact. For the PF-based diets FO continued to be the main environmental impact contributor, presenting however a smaller contribution. The replacement by SBO presented similar results, being again the FM from Peruvian anchovy the main contributor to environmental impact for the FM-based diets. While for PF- Diets 100% fishmeal (FM) 80% plant feedstuffs (PF) 20% fishmeal (FM) P/CH ratio P40/CH20 FM P40/CH20 P50/CH10 FM P50/CH10 P40/CH20 PF P40/CH20 P50/CH10 PF P50/CH10 S1—aquafeed ingredients FM from Peruvian anchovy Fish oil from Peruvian anchovy Soybean meal Maize gluten 95.0 34.8 23.5 – – 96.2 45.2 21.3 – – 94.7 8.6 38.1 17.0 2.3 96.1 11.5 39.6 23.8 2.8 Wheat gluten Pregelatinized maize starch Cellulose Monocalcium phosphate All transportations until MZS S2—processing unit Mixing Pelletization Total single score (Pt) – 12.8 28.9 – 3.1 1.8 39.4 60.6 0.235 – 6.6 26.8 – 1.9 1.9 39.4 60.6 0.223 10.0 13.1 8.1 2.8 3.1 2.3 39.4 60.6 0.191 15.1 5.0 – 2.1 1.4 2.5 39.4 60.6 0.175 Results reported to the functional unit: 1 kg of experimental diet Int J Life Cycle Assess Table 6 Environmental contribution (%) by stage to each experimental diet and the total single score, after replacing fish oil (FO) from by-products by 90% of soybean oil (SBO) and 10% FO from Peruvian anchovy, using Europe Recipe H/A V1.13 method (P: protein; CH: carbohydrates; MZS: Experimental Marine Zoology Station) Protein source Diets 100% fishmeal (FM) 80% plant feedstuff (PF) 20% fishmeal (FM) P/CH ratio P40/CH20 FM P40/CH20 P50/CH10 FM P50/CH10 P40/CH20 PF P40/CH20 P50/CH10 PF P50/CH10 S1—aquafeed ingredients FM from Peruvian anchovy 94.7 37.2 95.9 48.0 94.1 9.6 95.7 12.9 FO from Peruvian anchovy 2.5 2.3 4.3 4.4 Soybean oil Soybean meal Maize gluten 15.7 – – 14.2 – – 26.7 18.9 2.6 27.8 26.7 3.1 Wheat gluten – – 11.2 17.0 Pregelatinized maize starch 13.7 7.1 14.6 5.6 Cellulose 30.9 28.5 9.1 – Monocalcium phosphate All transportations until MZS S2—processing unit – 3.3 2.0 – 2.0 2.1 3.1 3.4 2.5 2.4 1.6 2.8 Mixing Pelletization Total single score (Pt) 39.4 60.6 0.220 39.4 60.6 0.210 39.4 60.6 0.172 39.4 60.6 0.157 Results reported to the functional unit: 1 kg of experimental diet based diets, the mixture of 90% of SBO plus 10% of FO from Peruvian anchovy contributed with the highest value for the environmental impact. This mixture contributed with 16.5% for diet FM P50/CH10 and 32.2% for diet PF P50/CH10. As stated by García et al. (2016), the difference in environmental impact between FM-based and PF-based diets can be explained by the overall lower environmental impact of PF ingredients, as wheat gluten and maize gluten. Accordingly, in both hypothetical scenarios for PF-based diets, SBM was the protein source with the largest environmental impact contribution, while glutens (wheat and maize) were the protein sources with the lowest contribution. However, Boissy et al. (2011) further concluded that the environmental impact of feeds depended highly on the geographic origins of fishery ingredients (e.g. FO from Norway or Peru) and of the type of agricultural crops used as feed ingredients (e.g. palm oil or rapeseed oil). LCA results from literature available to analyze and compare aquafeeds production is scarce and often controversial. For instance, Boissy et al. (2011) used the CML2 methodology to compare a standard diet with a low fishery product diet (LFD) for Atlantic salmon and rainbow trout. Both diets included fisheries and PF ingredients, but in different proportions. Results showed that LFD, which had higher PF content, had the highest contribution to most of the impact categories. However, when the environmental impact was assessed for the production of 1 tonne of fish, no differences were found between diets. On the contrary, Samuel-Fitwi et al. (2013) used the same impact assessment methodology (CML2) and concluded that the environmental impact of producing 1 tonne of FM-based diet for trout was higher than that of a diet where FM was replaced by soybean and rapeseed meal. This study used pelletization as the manufacturing process, while all the above studies used extruded diets. The manufacturing process may change the resulting environmental impacts of aquafeeds. However, to the authors’ knowledge, there is any study available comparing the environmental impact of pelletization versus extrusion. The controversial available literature results on the environmental impact to fish diets might be explained by several reasons. At first, different species have different nutritional requirements (NRC 2011), leading to the need of different aquafeeds formulations. Second, ingredients included in aquafeeds have different characteristics and properties, such as different sources (e.g. Portugal, Brazil, Peru), different manufacturing processes, and different nutritional quality (protein and lipid content). All these aspects influence the environmental impacts associated with diets or even aquaculture production. 4 Conclusions, limitations and future research This study assessed the environmental impacts of producing four experimental diets to gilthead seabream with different Int J Life Cycle Assess dietary protein sources and different P/CH ratios by applying the Recipe impact assessment methodology. Results of the present study showed that all diets tested had the same order of magnitude for the environmental impact score, independently of dietary protein source or P/CH ratio. S1 was the system with the largest contribution to the overall environmental impact. Fisheries-derived ingredients were the highest contributors to the environmental impact. Moreover, it was shown that replacing FO from fisheries byproducts by FO from Peruvian anchovy fishery or by SBO lead to a reduction of the environmental impact of all diets. Those scenarios highlight a tendency for PF-based diets having lower environmental impact scores than FM-based diets. One limitation of the present study was related with diet production, as pelletization was used while extrusion is the most current procedure in industry. In addition, the benefits brought by the valorization of fish by-products in producing FM, as alternative to their potential rejection to the environment, were not accounted in the impact analysis. Future research will have crucial importance to extend the boundaries of the present study to gilthead seabream production. Covering the full life cycle stages of aquaculture production, it will allow to provide solid background for more sustainable aquaculture. Finally, this study allowed to strengthen conclusions on the influence that certain parameters have in the resulting environmental impacts of aquafeeds, such as species of fed fish, since different species have different nutritional requirements; aquafeed ingredients origin; nutritional quality of aquafeed ingredients; conditions and manufacturing processes applied to raw material production; and aquafeeds production (pelletization or extrusion). The influence and effects of each of the previously mentioned parameters needs to be further explored and understood, particularly regarding ingredients since ingredients selection appears to be crucial for the reduction of environmental impact. Conflict of interest The authors declare that they have no conflict of interest. Funding information Catarina Basto-Silva and Inês Guerreiro were supported by FCT (Foundation for Science and Technology) grants (SFRH/BD/130171/2017 and SFRH/BPD/114959/2016, respectively). References Abdou K, Aubin J, Romdhane MS, Le Loc’h F, Lasram FBR (2017) Environmental assessment of seabass (Dicentrarchus labrax) and seabream (Sparus aurata) farming from a life cycle perspective: a case study of a Tunisian aquaculture farm. Aquaculture 471:204– 212 Achten WMJ, Acker K (2015) EU-average impacts of wheat production: a meta-analysis of life cycle assessments. J Ind Ecol 20:132–144 Almeida C, Vaz S, Ziegler F (2015) Environmental life cycle assessment of a canned sardine product from Portugal. J Ind Ecol 19(4):607–617 AOAC (2000) Official methods of analysis of AOAC. Association of Official Analytical Chemists, Gaithersburg Aubin J (2013) Life cycle assessment as applied to environmental choices regarding farmed or wild-caught fish. CAB Reviews 8(011):1-10 Avadí A, Henriksson PJG, Vázquez-Rowe I, Ziegler F (2018) Towards improved practices in life cycle assessment of seafood and other aquatic products. Int J Life Cycle Assess 23:979–981 Avadí A, Pelletier N, Aubin J, Ralite S, Núñez J, Fréon P (2015) Comparative environmental performance of artisanal and commercial feed use in Peruvian freshwater aquaculture. Aquaculture 435:52–66 Beutler (1984) Starch. In: Bergmeyer HU (ed) Methods of enzymatic analysis vol. 6. Verlag Chemie, Weinheim, pp 2–10 Boissy J, Aubin J, Drissi A, van der Werf HMG, Bell GJ, Kaushik SJ (2011) Environmental impacts of plant-based salmonid diets at feed and farm scales. Aquaculture 321:61–70 Carvalho A, Mimoso AF, Mendes AN, Matos HA (2014) From a literature review to a framework for environmental process impact assessment index. J Clean Prod 64:36–62 Cavalett O (2008) Análise do ciclo de vida da soja. Universidade Estadual de Campinas, Campinas Cavalett O, Ortega E (2009) Emergy, nutrients balance, and economic assessment of soybean production and industrialization in Brazil. J Clean Prod 17:762–771 Deng Y, Achten WMJ, Van Acker K, Duflou JR (2013) Life cycle assessment of wheat gluten powder and derived packaging film. Biofuels Bioprod Biorefin 7:429–458 Dias J, Conceição LEC, Ribeiro AR, Borges P, Valente LMP, Dinis MT (2009) Practical diet with low fish-derived protein is able to sustain growth performance in gilthead seabream (Sparus aurata) during the grow-out phase. Aquaculture 293:255–262 Edwards P (2015) Aquaculture environment interactions: past, present and likely future trends. Aquaculture 447:2–14 Enes P, Panserat S, Kaushik S, Oliva-Teles A (2008) Growth performance and metabolic utilization of diets with native and waxy maize starch by gilthead sea bream (Sparus aurata) juveniles. Aquaculture 274(1):101–108 Fantin V, Righi S, Rondini I, Masoni P (2017) Environmental assessment of wheat and maize production in an Italian farmers’ cooperative. J Clean Prod 140:631–643 FAO (2016) The state of world fisheries and aquaculture: opportunities and challenges. Food and Agriculture Organization of the United Nations, Rome 243 p FAO (2018) The state of world fisheries and aquaculture: meeting the sustainable development goals. Food and Agriculture Organization of the United Nations, Rome, 227 pp FAO (2019) Fisheries and aquaculture software. FishStatJ—software for fishery statistical time series. In: FAO Fisheries and Aquaculture Department [online]. Rome. Updated 21 July 2016. http://www. fao.org/fishery/. Accessed 25/02/2019 Fréon P, Avadí A, Vinatea Chavez RA, Ahón FI (2014) Life cycle assessment of the Peruvian industrial anchoveta fleet: boundary setting in life cycle inventory analyses of complex and plural means of production. Int J Life Cycle Assess 19(5):1068–1086 Fréon P, Durand H, Avadí A, Huaranca S, Moreyra RO (2017) Life cycle assessment of three Peruvian fishmeal plants: toward a cleaner production. J Clean Prod 145:50–63 García BG, Jiménez CR, Aguado-Giménez F, García JG (2016) Life cycle assessment of gilthead seabream (Sparus aurata) production in offshore fish farms. Sustainability 8:1228–1247 Gundoshmian TM, Ershadi SZ, Ebrahimi S (2017) Life cycle assessment of wheat production by using LCA methodology in Ardabil, Iran. Foreign Agricultural Economic Report ISSN: 0429-0577 101–108 Henriksson PJG, Mohan CV, Phillips MJ (2017) Evaluation of different aquaculture feed ingredients in Indonesia using life cycle assessment. IJoLCAS 1:13–21 Int J Life Cycle Assess Hofstetter P (1998) Perspectives in life cycle impact assessment—a structured approach to combine models of the technosphere, ecosphere and valuesphere. Kluwer Academic Publishers, Dordrecht IEA (2016) Electricity heat in Portugal, available from the International Energy Agency IFFO (2016) IFFO: fishmeal and fish oil statistical yearbook Iribarren D, Moreira MT, Feijoo G (2012) Life cycle assessment of aquaculture feed and application to the turbot sector international. Int J Environ Res 6(4):837–848 ISO 14040 (2006) Environmental management—life cycle assessment— principles and framework. International organisation for standardisation (ISO), Geneve ISO 14044 (2006) Environmental management—life cycle assessment— requirements and guidelines. International organisation for standardisation (ISO), Geneve Izquierdo MS, Turkmen S, Montero D, Zamorano MJ, Afonso JM, Karalazos V, Fernández-Palacios H (2015) Nutritional programming through broodstock diets to improve utilization of very low fishmeal and fish oil diets in gilthead sea bream. Aquaculture 449:18–26 MapQuest (2018) Official MapQuest http://www.mapquest.com/. Accessed 14 august 2018 Martínez-Llorens S, Vidal AT, Moñino AV, Torres MP, Cerdá MJ (2007) Effects of dietary soybean oil concentration on growth, nutrient utilization and muscle fatty acid composition of gilthead sea bream (Sparus aurata L.). Aquac Res 38:76–81 Montero D, Grasso V, Izquierdo MS, Ganga R, Real F, Tort L, Caballero MJ, Acosta F (2008) Total substitution of fish oil by vegetable oils in gilthead sea bream (Sparus aurata) diets: effects on hepatic Mx expression and some immune parameters. Fish Shellfish Immunol 24(2):147–155 Naylor RL (2016) Oil crops, aquaculture, and the rising role of demand: a fresh perspective on food security. Glob Food Sec 11:17–25 NRC (2011) Nutrient requirements of fish and shrimp. The National Academies Press, Washington, DC Olsen RL, Hasan MR (2012) A limited supply of fishmeal: impact on future increases in global aquaculture production. Trends Food Sci Technol 27:120–128 Pelletier N, Tyedmers P (2007) Feeding farmed salmon: is organic better? Aquaculture 272:399–416 Robaina L, Izquierdo MS, Moyano FJ, Socorro J, Vergara JM, Montero D, Fernández-Palacios H (1995) Soybean and lupin seed meals as protein sources in diets for gilthead seabream (Sparus aurata): nutritional and histological implications. Aquaculture 130:219–233 Rust MB, Barrows FT, Hardy RW, Lazur A, Naughten K, Silverstein J (2011) The future of aquafeeds vol 124. NOAA/USDA, Silver Spring Samuel-Fitwi B, Meyer S, Reckmann K, Schroeder JP, Schulz C (2013) Aspiring for environmentally conscious aquafeed: comparative LCA of aquafeed manufacturing using different protein sources. J Clean Prod 52:225–233 Silva CB, Valente LMP, Matos E, Brandão M, Neto B (2018) Life cycle assessment of aquafeed ingredients. Int J Life Cycle Assess 23:995– 1017 Strazza C, Magrassi F, Gallo M, Borghi AD (2015) Life cycle assessment from food to food: a case study of circular economy from cruise ships to aquaculture. Sustainable Production and Consumption 2: 40–51 Tacon AGJ, Metian M (2008) Global overview on the use of fish meal and fish oil in industrially compounded aquafeeds: trends and future prospects. Aquaculture 285(1–4):146–158 Tacon AGJ, Metian M (2015) Feed matters: satisfying the feed demand of aquaculture. Rev Fish Sci Aquac 23:1–10 Tacon AGJ, Metian M (2018) Food matters: fish, income, and food supply—a comparative analysis. Rev Fish Sci Aquat 26:15–28 Van Stappen F, Loriers A, Mathot M, Planchon V, Stilmant D, Debode F (2015) Organic versus conventional farming: the case of wheat production in Wallonia (Belgium). Agric Agric Sci Procedia 7:272–279 Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.