IJFS#1077_bozza Ital. J. Food Sci., vol. 30, 2018 - 362 PAPER THE ECO-EFFICIENCY OF THE DAIRY CHEESE CHAIN: AN ITALIAN CASE STUDY M.B. FORLEO*a, N. PALMIERIa and E. SALIMEIb aDepartment of Economics, University of Molise, Via F. De Sanctis, 86100 Campobasso, Italy bDepartment of Agriculture, Environment and Food Sciences, University of Molise, Via F. De Sanctis, 86100 Campobasso, Italy *Corresponding author: Tel. +39 0874404454 *E-mail address: forleo@unimol.it ABSTRACT The eco-efficiency of mozzarella cheese production was investigated in two dairy chains that differ in liquid whey recycling, with whey recycling (B) and without whey recycling (A), in cow diets. The total eco-efficiency (total GVA/total GWP) for 1 kg of mozzarella cheese ranged from € 0.19 (B) to € 0.16 per kg CO2-eq (A). The cheese-making phase of each diet accounted for about 3% of GWP total emissions. The mozzarella cheese making phase had the highest eco-efficiency ratio, while the milk production phase showed the lowest economic value and the highest impact. Findings suggest improvements in reducing the environmental burden of the primary phase while increasing its economic value. Keywords: carbon footprint, cheese whey recycling, eco-efficiency ratio; economic value added, mozzarella cheese production Ital. J. Food Sci., vol. 30, 2018 - 363 1. INTRODUCTION Food supply chains are increasingly associated with environmental impacts, and this has brought global attention to the sustainability of the agri-food systems (FANTOZZI et al., 2015). Dairy products have a great impact, especially in terms of resource depletion and greenhouse gas emissions (GONZÁLEZ-GARCÍA et al., 2013). Furthermore, the dairy industry is considered responsible for a significant impact due to the characteristics of its wastewaters and effluents (MIRABELLA et al., 2014). Solid waste treatment and wastewater treatment along the dairy chain affect several environmental indicators. Cheese whey is the main pollutant generated from cheese production that can cause several environmental impacts (PRAZERES et al., 2012). Thus, cheese whey cannot be discharged directly into the environment without appropriate treatment. According to some authors (SUCCI et al., 1986), apart from potential environmental benefits, liquid whey is also an interesting animal diet ingredient from an economic point of view, especially when distances from the cheese industry are short and costs of handling and transportation are high. In the framework of a circular economy approach, the reuse of whey in dairy cows' diet may minimize resources use and waste production from cheese making. In this regard, the European Commission has recently adopted an action plan on the circular economy - where the value of products, materials, and resources is maintained in the economy as long as possible, and the generation of waste is minimized- to develop a sustainable and competitive economy with low carbon content and efficient resource use. Assessing the environmental performance of dairy chains can reduce their impacts and improve the efficiency of resource use (MU et al., 2017). Life cycle assessment (LCA) is a methodology widely used to investigate the environmental impact of food production. SALA et al. (2017) underlined the importance of the environmental and socio-economic impacts associated with the food supply chains and indicated life cycle thinking and assessment as key elements in identifying more sustainable solutions for global food challenges. Furthermore, NOTARNICOLA et al. (2015) deepened the issue of LCA in the agri-food sector with case studies, methodological issues and best practices. Existing literature reports several studies that addressed different topics related to the LCA of cheese production. KIM et al. (2013) conducted a US-based LCA to determine the environmental impacts of cheddar, mozzarella cheese and dry whey from cradle-to-grave. GONZÁLEZ-GARCÍA et al. (2013) studied the life cycle of mature cheese production in Portugal from a cradle-to-gate perspective and identified the environmental hotspots. PALMIERI et al. (2017) applied an LCA approach to assess the impacts of mozzarella cheese production and evaluate the contribution of different strategies in a traditional dairy chain. Global warming potential is one of the most studied impacts of dairy products. ROTZ (2018) reviewed the models for evaluating GHG emission from dairy farms —along a continuum from relatively simple models for single GHG emission sources to very detailed simulations over the whole farm production system— and concluded that LCA is a comprehensive method for quantifying and evaluating the different sources of emissions over the full cycle. COLOMBINI et al. (2015) applied an LCA cradle-to-farm-gate to assess the global warming potential of milk production in three forage systems scenarios and lactating cow diets. HAWKINS et al. (2015) estimated how the formulation of the ration and the associated land allocation decisions, contribute to reductions in GHG emissions of the intensive dairy production systems in Ontario. VAN MIDDELAAR et al. (2013) studied the environmental effect of replacing grass silage with maize silage in a feeding strategy Ital. J. Food Sci., vol. 30, 2018 - 364 and applied a life cycle assessment to predict GHG emissions at chain level. Finally, FINNEGAN et al. (2015) measured the global warming potential associated with the processing of raw milk into 11 dairy products in the Republic of Ireland following a cradle-to-processing factory gate boundary. A general result from literature suggested that raw milk production is the most impactful phase along the chain due to feed production and animal emissions. Few studies dealt specifically with the environmental impact of mozzarella cheese production. Two studies investigated the impact of American and Canadian mozzarella cheese production (KIM et al., 2013; VERGÉ et al., 2013) by considering several impact categories. Concerning the Italian mozzarella product, a study (DALLA RIVA et al., 2017) investigated a cradle-to-processing-gate LCA of two types of mozzarella (the traditional one produced from raw milk, and the mozzarella obtained from curd) focusing mainly on transformation and consumption of mozzarella cheese, also dealing with different environmental impacts. A study by PALMIERI et al. (2017) focused on several impact categories of both farm and factory phases based on some study cases of the mozzarella production in Italy. HELMES et al. (2016) assessed the carbon footprint of an Italian mozzarella facility dealing with the sensitivity of LCA results according to different allocation choices. Finally, FALCONE et al. (2017) applied the LCA approach to assess the environmental effect of a shelf life extension technique in the lacto fermented Italian mozzarella cheese production. Under a wider sustainable perspective, the assessment of a dairy product should be extended beyond environmental impacts by considering its profitability and economic performance. Recent studies started focused on the economic and environmental assessment of dairy products by using different approaches and focusing on minimising costs and/or on maximising profits. SOTERIADES et al. (2016) proposed to combine the LCA approach with the Data Envelopment Analysis (DEA) method in order to holistically assess dairy farm eco- efficiency by maximising output per unit of environmental impacts. KIRILOVA and VAKLIEVA-BANCHEVA (2017) designed an optimal “green” portfolio for curd production in Bulgaria to demonstrate the role of the environmental impacts - measured in terms of wastewater and CO2 emissions- within a profit maximization function that includes the costs of the above impacts. MURPHY et al. (2017) compared male dairy calf-to-beef production systems based on different animal performance and applied economic profitability and GHG emissions models to highlight the best performing system per each perspective. HAWKINS et al. (2015) used an optimization model of ration formulation to determine how specific GHG targets can be reached while maximising net returns to an intensive dairy farming system. WETTEMANN and LATACZ-LOHMANN (2017) estimated the potential costs and GHG emissions savings for a sample of 216 dairy farms in northern Germany using an input- oriented Data Envelopment Analysis and showed that cost and GHG emission reductions are complementary across a wide range. An economic approach focused on costs is also followed by HUYSVELD et al. (2017) that analysed a sample of 103 specialized dairy farms in Flanders (Belgium) and showed potential simultaneous savings in costs and overall natural resource demand (up to 48%). FALCONE et al. (2017) applied a Life Cycle Assessment and Life Cycle Costing methods in order to assess the environmental and economic impacts of innovations in the Lacto-fermented mozzarella cheese production in Calabria region. Finally, HESSLE et al. (2017) studied different production scenarios of the dairy chain in Sweden by performing a Life Cycle method to assess the best environmental performance and by quantifying the costs in the primary production of dairy and beef to find out the most cost-efficient production models. Ital. J. Food Sci., vol. 30, 2018 - 365 Another approach that integrates economic and environmental assessment is based on the eco-efficiency ratio (SALING, 2016). Eco-efficiency is defined as economic efficiency combined with environmental benefits and deals with three main goals: the reduction of resource consumption, the reduction of environmental impacts, and the increase of product value. The concept of eco-efficiency has been applied to several agricultural products to estimate the value added per kg of GHG emitted into the atmosphere for each system studied. In the dairy sector, BASSET-MENS et al. (2009) applied an eco-efficiency analysis of milk production in Flanders. MEUL et al. (2007) studied the eco-efficiency of milk production in some Flemish dairy farms, but the authors intended eco-efficiency in terms of ecologic and not economic terms and measured an indicator based on nitrogen and energy use efficiency. To the best of our knowledge, few studies considered the eco-efficiency of the dairy chain. A study measured the economic performance of the cheese production chain by calculating the gross value added (GVA) of stages along the chain (VAN MIDDELAAR et al., 2011). Another study (SANJUAN et al.. 2011) measured the economic added value and the net income of Mahon-Menorca cheese production under different scenarios regarding technical and cleaner production criteria. However, that study included the assessment of the cheese production phase and excluded the milk production phase. A different approach to eco-efficiency was applied in a study that related the environmental performances with the economic efficiency in the use of dairy farms inputs (IRIBARREN et al., 2011). This study aims to contribute to the literature on the environmental and economic performances of the mozzarella cheese production by measuring its eco-efficiency ratio based on an Italian case study. The study answers the question, "how much value is added per kg of GHG emitted to the atmosphere?". Firstly, the environmental and economic assessments were implemented; subsequently, the two perspectives were combined within an eco-efficiency analysis. In an earlier study (PALMIERI et al., 2017) an environmental analysis was performed according to a global approach. The present study goes further by focusing on the carbon footprint assessment and adding the analysis of the economic performance of mozzarella cheese production. 2. MATERIALS AND METHODS 2.1. The environmental assessment 2.1.1 Goal and scope definition The main purpose of the study was to calculate the eco-efficiency ratio of mozzarella cheese production based on raw milk produced following different feeding strategies. The environmental impact of the dairy cheese chain was based on GHG emissions, and the economic performances considered the GVA of the dairy cheese chain. The value added per GHG emission of one kg of mozzarella cheese produced was finally measured. The carbon footprint (CF), an important index of the climate change impacts of food production within the whole supply chain (ROMA et al., 2015), was measured by an Attributional Life Cycle Assessment methodology (BAITZ, 2017; ISO 14040, 2006; ISO 14044, 2006). The CF of 1 kg of mozzarella cheese is defined as the sum of all GHGs emitted along the production cycle (RÖÖS et al., 2014). GWP is expressed in CO2 equivalent (CO2-eq) using weights of 1,28 and 265 for carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O), respectively (assuming 100 years lifespan; IPCC, 2015). Ital. J. Food Sci., vol. 30, 2018 - 366 Furthermore, an economic analysis considered the added economic value of the dairy cheese chain as the difference between total revenues and total costs for intermediate consumption (VAN MIDDELAAR et al., 2011). Intermediate consumption costs measure the value of goods and services consumed, including raw materials, services, and other operating expenses, other than fixed assets. The GVA does not include labor costs, depreciation, nor interest loan payment; when considering the depreciation of fixed capital, a net value added is obtained. The GVA indicator was chosen because it is frequently used to measure the economic sustainability of agricultural systems (VAN MIDDELAAR et al., 2011). The final goal of jointly assessing the environmental and economic performances in the case study was pursued by measuring the eco-efficiency ratio (GVA/GWP) of mozzarella cheese production based on milk produced following different feeding strategies. 2.1.2 Functional unit and system boundary The functional unit (FU) of the environmental and economic analysis was expressed per 1 kg of mozzarella cheese produced from 8.11 L of cow milk. The LCA system boundary (Fig. 1) refers to the first two phases of a dairy chain, namely the dairy farming and the cheese-making phases. The boundary considers: the dairy farm - including the agricultural processes of feedstuffs and the whole life cycle of cows -; and the cheese factory -including all the activities that take place for the mozzarella cheese making, from the milk reception to the mozzarella production and the whole liquid whey disposal (the wastewater treatment plant or recycled into the cow diets). Two dairy diets that differ in the usage/non-usage of liquid whey were assessed. In relation to the different disposal of the liquid whey, along with the two diets (A and B diets), two different chains are considered. In A chain, the whole amount of liquid cheese whey is mixed with the wastewater effluent from the mill and delivered to a municipal wastewater treatment plant (Fig. 1). In the B chain, the whole amount of liquid whey produced at cheese-making level is delivered to the farm where it is used, after microbial stabilization, in animal feeding as partial substitute of drinking water. The physical allocation method was used in the baseline scenario to share the environmental burden between milk and meat at the farm level, while the environmental burden of the mozzarella production was totally allocated to curd (GONZÁLEZ-GARCÍA et al., 2013). The percentages of physical allocation at case farm level were 88% to milk and 12% to meat (as live weight cow and calf) (IDF, 2015). The manure/slurry allocation was not necessary because farmyard manure was recycled as fertilizer in the feed cultivation. Ital. J. Food Sci., vol. 30, 2018 - 367 Figure 1. System boundaries: dairy farm and cheese factory. Abbreviations: See Table 1. Table 1. List of Abbreviations and Acronyms. ALCA Attributional Life Cycle Assessment CF Carbon Footprint CO2 Carbon dioxide CU Cereal Unit allocation method CH4 Methane FPCM Fat and Protein Corrected Milk FU Functional Unit GVA Gross Value Added GHG Greenhouse Gas Emissions GWP Global Warming Potential IPCC Intergovernmental Panel on Climate LCA Life Cycle Assessment N2O Nitrous Oxide WWTP Wastewater treatment plant Ital. J. Food Sci., vol. 30, 2018 - 368 2.1.3 Life cycle inventory Data for the life cycle inventory analysis partly comes from the INLATTE Project (Tables 2-4) and were collected through a questionnaire drawn according to the guidelines for the application of LCA to food and agricultural products (NERI, 2009). Secondary data (Table 5) were taken from both the ECOINVENT database v. 3.0 (WEIDEMA et al., 2013) and literature (FRANCHINI and NERI, 2004; NERI and BORSARI, 2005; KIM et al., 2013). Primary data were collected from two firms (a dairy farm and a cheese factory) located in Molise region (IT). Data from the case farm reported the milk quantity and quality, the Italian Friesian cow rations and water consumption, and the manure/slurry produced. The case farm experimented two different dietary strategies: a diet including ensiled forages and no liquid whey usage (A diet) and a diet including both silages and liquid whey (B diet). Data reported in Table 2 summarise the management of animals in the case farm. For the present study, 36 lactating cows were divided into two groups of 18 cows each which were homogeneous and comparable in terms of milk yield and days of lactation and parity. The average fat and protein corrected milk (FPCM) yield has been calculated on a 305 days basis for each experimental group and used in the LCA study. The FPCM yield was calculated according to FINNEGAN et al. (2015). Table 3 shows the composition of the diets. In this regard, it is worth noting that feedstuffs were offered as total mixed rations, except for the microbiologically stabilized liquid cheese whey offered to B diet cows as partial substitute of drinking water. Water consumption in B diet was, therefore, lower than that in the A diet. Primary data from the cheese factory have been recorded throughout the experiment and summarised in Table 4. Mozzarella cheese for fresh consumption traditionally obtained directly and solely from liquid milk is the dairy product considered in the study. Table 2. Case farm characteristics. Case farm data Cow breed Holstein Friesian Number of lactating cows 36 Number of dry cows 9 Dairy replacement calves and heifers, n. 32 Number of calves (male) 18 Days of production/year (lactating cows) 305 Males raised as beef cattle, age (days) Calves: 20 Milk production Diets Milk yield – FPCM (kg/per yr) A 8,332 B 8,039 % Fat A 4.03 B 3.99 % True Protein A 3.68 B 3.60 Ital. J. Food Sci., vol. 30, 2018 - 369 Table 3. Water consumption and characteristics of diets on a dry matter (DM) basis. Diets Calves diet A B Water consumption (L/day) 10 10 Liquid whey (kg/day) - - Total DM intake (kg/ day) 1.96 1.96 Heifers diet A B Water consumption (L/day) 35 25 Liquid whey (kg/day) - 0.57 Total DM intake (kg/ day) 4.53 5.10 Lactating cow diet A B Water consumption (L/day) 80 50 Liquid whey (kg/day) - 1.48 Total DM intake (kg/ day) 20.06 21.54 Dry cow diet A B Water consumption (L/day) 40 40 Liquid whey (kg/day) - - Total DM intake (kg/ day) 13.08 13.08 When real data were not available, inventory data were collected from literature and ECOINVENT database (v. 3.0) (WEIDEMA et al., 2013), as reported in Table 5. Emissions considered in the study were drawn from literature (Table 6). Data for the raw milk and whey transportation and for the wastewater treatment plant for whey disposal came from ECOINVENT database. Table 4. Cheese factory data. Products data kg of mozzarella produced by 8.11 L of milk 1 kg of whey produced by 1 kg of mozzarella 0,89 Fat in mozzarella (g/kg of product) 185 Protein in mozzarella (g/kg of product) 154 Fat in whey (g/kg of product) 2 Protein in whey (g/kg of product) 7 Resources consumption Electricity consumption (kWh/ kg of mozzarella) 0,20 Heat consumption (MJ/kg of mozzarella) 0,11 Water consumption (L/kg of mozzarella) 18,08 Data source: INLATTE Project. Ital. J. Food Sci., vol. 30, 2018 - 370 Table 5. Secondary data considered in the study. Source Feed cultivation and processing Barley ECOINVENT DATABASE (v. 3.0) Maize Meadow hay Milk powdered FRANCHINI and NERI (2004); ECOINVENT DATABASE (v. 3.0) Mixed feed ECOINVENT DATABASE (v. 3.0) Mineral feed Sugar beet pulp Soybean meal 44% Triticale silage Mozzarella production Milk reception ECOINVENT DATABASE (v. 3.0); Pasteurisation FRANCHINI and NERI (2004); ECOINVENT DATABASE (v. 3.0) Heating, inoculation and coagulation ECOINVENT DATABASE (v. 3.0) Curd cutting Curd transfer and ripening Spinning and molding Hardening and salting Raw milk transportation ECOINVENT DATABASE (v. 3.0) for diesel track of 16 t capacity. Real distance from the dairy farm to the factory 10 km Wastewater treatment ECOINVENT DATABASE (v. 3.0); moderately large municipal wastewater treatment plant with a three-stage process (mechanical, biological and chemical) Table 6. Emissions considered in the study. Emissions Source Enteric and animal housing emissions CH4 emissions and the ammonia emissions BATTINI et al. (2016); EMEP/EEA (2009) Nitrous oxide (N2O) emissions from animal housing Not considered according to BATTINI et al. (2016) Storage emissions Emissions of methane (CH4) and nitrous oxide (N2O) DALLA RIVA et al. (2014); IPCC (2006) (Tier 2); using ISPRA (2008) methods Ammonia (NH3) emissions due to manure/slurry storage FALCONI et al. (2011) using ISPRA (2008) method Nitrogen oxides (NOx) emissions BATTINI et al. (2016) using the factor by IPCC (2006) Emissions related to manure/slurry spreading N2O, NH3, NOx and nitrate leaching BATTINI et al. (2016) using IPCC (2006) The P leaching run-off emissions BATTINI et al. (2016) Emission factor of Potassium, Copper and Zinc NERI and BORSARI (2005) 2.2. Economic assessment and eco-efficiency ratio of the dairy chain The eco-efficiency indicator is based on data from both environmental and economic accounting systems. The higher the indicator value, the higher the economic performance per unit of environmental burden. Since ecological and economic data need to be derived Ital. J. Food Sci., vol. 30, 2018 - 371 from the same data set (MULLER et al., 2015), we collected information based on the annual budget of the considered dairy farm and the cheese factory. The economic data for both stages, milk production and mozzarella cheese making, are shown in Table 7. The B dairy chain had lower total costs than A chain due to both the elimination of treatment costs of whey in the WWTP and saved transportation costs of whey from the cheese factory to the dairy farm. The factory and the farm agreed to equally share the costs of both whey transportation (from the cheese factory to the dairy farm) and whey management at firm’s level. Finally, the lower costs of B chain were due to the reduction of water consumption in the diet. The eco-efficiency analysis was applied to the two stages of mozzarella cheese production (i.e., milk production and mozzarella cheese-making phases). The eco-efficiency of each stage was computed by dividing its economic value added by its ecological impact (VAN MIDDELAAR et al., 2011). Table 7. Cheese factory and dairy farm economic data. Economic data Units Cheese factory (€/kg of mozzarella) Dairy Farm (€/8,11 L of milk) A chain B chain A chain B chain Gross revenue €/kg 6.10 6.10 4.00 4.00 Variable and fixed costs €/kg 5.10 4.90 3.44 3.37 Economic value added €/kg 1.00 1.20 0.56 0.63 Source: Data came from the dairy farm and cheese factory case studies. 2.3. Sensitivity analysis: allocation method and variability of GVA The choice of the allocation procedure for agricultural co-products may affect the results of LCA study as discussed in FLYSJO et al. (2012) and HELMES et al. (2016). Both studies compared the dry matter and the economic allocation methods for assessing the impact of dairy industry and underlined the need for testing results against different approaches. For this reason, a sensitivity analysis for environmental impacts was performed by changing the allocation method of milk according to a cereal unit (CU) method (BRANKATSCHK and FINKBEINER, 2014). This sensitivity analysis involved only the case farm level, as in many reported studies (FANTIN et al., 2012; GONZÁLEZ-GARCÍA et al., 2013; KIM et al., 2013; VAN MIDDELAAR et al., 2011), because milk production is more impactful than cheese-making. The CU allocation method is based on the metabolizable energy content of product and co-product for feed purpose so that it allows considering agricultural products and co-products used in different sectors. The environmental burden was allocated 86.6% to milk, 6.8% to live-weight dairy cow and 6.6% to live-weight fattening male calf (BRANKATSCHK and FINKBEINER, 2014). Furthermore, if the economic dataset was based on the annual reports of the dairy farm and the cheese factory -and therefore are real and accurate-, a further sensitivity analysis was performed to estimate the effect a ±10% change of GVA of the two stages for each dairy chain. Ital. J. Food Sci., vol. 30, 2018 - 372 3. RESULTS AND DISCUSSION 3.1. The carbon footprint of 1 kg of mozzarella: baseline allocation Results of the environmental impact of 1 kg of mozzarella cheese showed that raw milk production was the most impactful phase along the considered supply chain, irrespective of the diet followed at the farm level (Fig. 2). Figure 2. Carbon footprint of 1 kg of mozzarella cheese in A supply chain (on the left side) and B chain (on the right side): milk and mozzarella production (physical allocation). Note: Transport refers both to the milk delivered to the dairy factory (supply chain A and B) and to the liquid whey delivered to the dairy factory (B supply chain) or the wastewater treatment plant (WWTP; A supply chain). Milk production was the most critical phase along the dairy chain, with contributions of 96% (A diet) and 97% (B diet) of the global warming potential (GWP). The high contribution of milk production phase to the environmental impact of the mozzarella dairy chain observed is consistent with the study of DALLA RIVA et al. (2017), even considering the farm gate-to grave perspective followed by the authors. A similar conclusion was in the study of FINNEGAN et al. (2015) that, although was based on different cheese product and fluid milk, showed that milk production contributes to GWP within 81% - 97% range (depending on the amount of raw milk per kg of the six cheese products considered in the study). The remainder contribution being mainly due to the processing phase. The environmental impacts of milk production phase were due to emissions of both methane from the enteric fermentation process and dinitrogen monoxide and carbon dioxide from manure management and spreading, confirming the study of GONZÁLEZ- GARCÍA et al. (2013) which referred to the cheese chain in Portugal. Methane from enteric fermentation and manure management was also the main GHG emission source in other studies dealing with cheese (KIM et al., 2013) and milk production (VIDA and TEDESCO, 2017). In the studies of VAN MIDDELAAR et al. (2011) and SANTOS et al. (2016), the enteric fermentation was the main emission source affecting GWP. According to VAN MIDDELAAR et al. (2011), the stage that contributed most to total global warming potential along the production chain of Dutch semi-hard cheese was on-farm milk production (65%), mainly due to enteric fermentation. In a study by SANTOS et al. (2016) about the cheese production in a small-sized dairy industry in Brazil, the contributions of the raw milk production ranged from 70 to 98% depending on the different midpoint impact categories. Ital. J. Food Sci., vol. 30, 2018 - 373 The cheese-making phase of each diet accounted for about 3% of GWP total emissions. Mozzarella production phase showed impacts due to carbon dioxide from heat consumption during the cheese making process. This result confirms VAN MIDDELAAR et al. (2011) findings that measured the contribution of semi-hard cheese-making and packaging phases in about 3% - 4% of GWP emissions, each. Even in the study of HELMES et al. (2016), the contribution from the processing step of mozzarella production was quite limited compared to raw milk and transport impacts. Furthermore, in our study, impacts of transportation of both milk —from dairy farm to cheese factory— and whey, either from factory to the wastewater plant or from factory to the dairy farm- were negligible due to the close distance between the locations of the two firms involved, the farm and the factory. A similar result was reported in the study of FINNEGAN et al. (2015) where liquid milk transportation contributed for less than 0.5%, whichever dairy products considered in the assessment. The relative burden of the wastewater treatment (in A diet) along the whole dairy chain was also considered insignificant. Comparing impacts between the chains, results based on a cradle-to-processing-gate boundary showed that the B dairy chain had a CF 1% higher than the A chain per unit of product. The carbon footprint of mozzarella cheese in A chain was 9.65 kg CO2-eq/kg mozzarella cheese, while it was 9.81 kg CO2-eq /kg mozzarella cheese in B chain. The B dairy chain, although with the liquid whey usage, appeared to be a slightly worse solution due to a lower milk yield (8,039 kg FPCM) compared with A chain (8,332 kg), confirming that the environmental impact increases at decreasing milk yields (NEMECEK et al., 2011). Study findings were similar to those reported in KIM et al. (2013) where the carbon footprint of US mozzarella cheese was 9.30 kg CO2-eq/kg. Furthermore, the results of our study are consistent with the study of HELMES et al. (2016), even if these authors considered different scenarios (mozzarella with ricotta or mozzarella with whey powder) from that of the present study. According to SANTOS et al. (2016), GWP emissions of cheese production were 14.44 kg CO2-eq/kg of product, while in VERGÉ et al. (2013) the carbon footprint of Canadian dairy products was significantly lower than the one assessed in this analysis. However, both studies cannot be directly compared to the present findings due to several differences related to the final cheese products, to the production process and different methodological choices. In our study, GWP emissions of mozzarella cheese-making phase were 0.32 kg CO2-eq with A diet and 0.29 with B diet. These findings are quite in line with the study of FINNEGAN et al. (2015) that calculated the GWP emission of six groups of dairy products (not mozzarella cheese) and showed that GWP emissions from the dairy processing phase ranged 0.11-2.5 kg CO2-eq/kg according to the different groups of studied products. In conclusion, despite different environmental assessment methods used in literature, the milk production is the process that mostly contributed to the environmental impact. Improvement alternatives at the dairy-farm level are therefore required, and they involve many aspects, among which is the use of fertilizers for feedstuffs cultivation. In this regard, KOESLING et al. (2017) assessed the variations in nitrogen utilisation of conventional and organic dairy farms in Norway. These researchers concluded that, for both a dairy farm and system area, N-surpluses increased with increasing use of fertilizer N per hectare, biological N-fixation, and imported concentrates and roughages, while they decreased with higher production per area. PAGANI et al. (2016) investigated direct and indirect energy inputs in a sample of dairy farms -either grain-based, forage-based or organic- and demonstrated that potential reduction in the overall energy input could be achieved by shifting to organic farming, switching to forage-based farming, and by promoting reduced use of fertilizers. Both studies highlighted the importance of good agronomy that utilizes available nitrogen and reduces energy inputs properly. Ital. J. Food Sci., vol. 30, 2018 - 374 Other studies focused on improvements in the composition of dairy ration to mitigate the environmental impact. HAWKINS et al. (2015) suggested that feeding decisions have important implications for GHG emissions from intensive dairy production due to the wide variation in emissions from alternative crops that can be used in the ration. PATRA et al. (2011) reviewed several potential methane mitigation options such as animal interventions (i.e., number and productivity of animals or genetic selection), dietary interventions, suppression of rumen methanogens, and new potential technologies, by underlying areas worthy of investigation for CH4 mitigation and improvements most likely to be adopted by farmers. Finally, WHITE (2016) proposed a farm-scale diet optimization model to reduce land use, water use, and GHG emissions within dairy production systems and assessed how improved energy and protein use efficiency reduces the environmental impacts of dairy production systems. Finally, improvements in the environmental profile of cheese production should also be directed at the dairy factory level, mainly due to a high-energy consumption of machinery used during the production process. However, according to VAN MIDDELAAR et al. (2013), mitigation strategies may be case-specific and must consider the level of the analysis –at animal, farm and chain level-. To achieve a sustainable mozzarella cheese production chain, not only its environmental impact must be considered and minimized, but also the economic value that is added along the chain. 3.2. The eco-efficiency of the dairy chain The total eco-efficiency (total GVA/total GWP) of 1 kg of mozzarella cheese accounted for € 0.19 per kg CO2-eq in the B supply chain and € 0.16 per kg CO2-eq in the A supply chain (Table 8). Findings showed that dairy chain in case of B diet had a better eco-efficiency ratio per unit of GHG emitted to the atmosphere. Table 8. Carbon footprint and gross value added (GVA) per functional unit (FU=1 kg mozzarella cheese), and eco-efficiency of the two stages in the dairy chain (Physical allocation). Stage GWP (kg CO2-eq/FU) Economic Performance GVA/FU (€) Eco-efficiency Total GVA/ total GWP A chain B chain A chain B chain A chain B chain Milk production 9.33 9.52 0.56 0.63 0.06 0.07 Mozzarella cheese- making 0.32 0.29 1.00 1.20 3.12 4.13 Total 9.65 9.81 1.56 1.83 0.16 0.19 Under the economic viewpoint, the B dairy chain had lower total costs than the A chain due to: 1) the elimination of treatment costs of whey in the WWTP at cheese factory level; 2) the reduction of water consumption due to whey usage in B diet; 3) finally, to lower transportation costs. The total value added for 1 kg of mozzarella cheese was € 1.56 for the A dairy chain and € 1.83 for the B chain. When considering the distribution of total GVA along the chain, milk production accounted for a lower economic weigh (36 % in A chain and 34 % in B chain) compared to the value contribution of the cheese making process. For the above reasons, mozzarella cheese making had the highest eco-efficiency ratio for each dairy chain (€ 3.12 Ital. J. Food Sci., vol. 30, 2018 - 375 in A chain and € 4.13 in B chain) and added the highest economic value per unit of environmental impact. The average GVA per 1 kg of fat and protein correct milk (FPCM) for the milk production phase was € 0.56 (per 8.11 kg FPCM to produce 1 kg of mozzarella) for the A dairy chain and € 0.63 per (8.11 kg FPCM to produce 1 kg of mozzarella) for the B chain. Our results were consistent with the VAN MIDDELAAR et al. (2011) study that calculated the economic performances of a cheese chain as defined in this study (i.e. gross value added per environmental impact of stages along a production chain) and showed that the milk production contributed 34% to the total GVA of mozzarella cheese production. Furthermore, the economic performance of mozzarella production phase accounted for € 1.00 for the A chain and for € 1.20 in the B supply chain, confirming the VAN MIDDELAAR et al. (2011) results that showed a GVA of € 1.04 for the cheese-making phase. The above differences, while negligible, were likely due to both different local markets, products, and manufacturing costs and prices. For this reason, two sensitivity analyses were carried out to test eco-efficiency results against changes in the economic indicator and to test environmental results against an allocation method different from the one applied in the baseline analysis. 3.3. Sensitivity analysis results Results of the sensitivity analysis confirm previous results about the eco-efficiency of mozzarella cheese production. The first sensitivity analysis (Table 9) showed that results from the CU allocation were lower than results achieved through a physical allocation for each dairy chain, but the differences in the value of the carbon footprint were negligible (around 1% for each chain). Furthermore, comparing findings based on CU allocation for the two dairy chains, results were consistent with those presented in Fig. 2 based on the physical allocation method (data are available on request). The B dairy chain confirmed its lower environmental performance. Table 9. Sensitivity results of the Carbon footprint to the allocation method (Physical and CU allocation). Stage GWP (kg CO2-eq/FU) Physical allocation CU allocation A chain B chain A chain B chain Milk production 9.33 9.52 9.18 9.37 Mozzarella cheese- making 0.32 0.29 0.32 0.29 Total 9.65 9.81 9.50 9.66 FU=1 kg mozzarella cheese The second sensitivity analysis (Table 10) was performed to estimate the effect of ±10% change of GVA for each stage, for each dairy chain and each allocation method on the eco- efficiency ratio. Compared with the baseline scenario, the ±10% change of GVA modified the eco-efficiency scores in the range ±0.04 €/kg CO2-eq, (e.g., from a score of 0.14 to 0.18 and from a score of 0.16 to 0.20 €/kg CO2-eq, respectively in the A and B chains under the physical allocation method). Finally, findings showed higher eco-efficiency values with a CU allocation than a physical allocation method. Even in this case, results reported small changes in the absolute values of the eco-efficiency per 1 kg of mozzarella cheese and Ital. J. Food Sci., vol. 30, 2018 - 376 showed that the best-performing dairy chains did not change. Therefore, the dairy chain in case of B diet had the best eco-efficiency ratio per unit of GHG emitted to the atmosphere. From the two sensitivity analysis, it is possible to affirm that study results are not very much influenced by the choice between the two considered allocation methods, nor by the change in the economic value added. Table 10. Sensitivity results of the Economic performance (±10% change of GVA) and of the Eco-efficiency scores in the two dairy chains (Physical versus CU allocation and ±10% change of GVA). Change Stage Economic* performance GVA/FU (€) Eco-efficiency* scores GWA/GWP (€/kg CO2-eq) Physical allocation CU allocation A chain B chain A chain B chain A chain B chain +10% of GVA Milk production 0.62 0.69 0.06 0.07 0.07 0.08 Mozzarella cheese-making 1.10 1.32 3.44 4.55 3.44 4.55 Total 1.72 2.01 0.18 0.20 0.19 0.21 Baseline scenario Milk production 0.56 0.63 0.06 0.07 0.06 0.07 Mozzarella cheese-making 1.00 1.20 3.12 4.13 3.13 4.14 Total 1.56 1.83 0.16 0.19 0.17 0.20 - 10% of GVA Milk production 0.50 0.57 0.05 0.06 0.05 0.06 Mozzarella cheese-making 0.90 1.08 2.81 3.72 2.81 3.72 Total 1.40 1.65 0.14 0.16 0.15 0.17 *The different allocation method (Physical or CU allocation) does not imply any variation in the economic performance (GVA), while it influences the environmental assessment (GWP, as reported in Table 9) and the eco-efficiency results (because the eco-efficiency is the ratio between Total GVA/total GWP). 4. CONCLUSIONS In this paper, the eco-efficiency ratio of mozzarella cheese production is assessed in an Italian case study according to the handmade cheese making system considering two different diets at the farm level, including or not including liquid cheese whey in cows’ diet. From an environmental point of view, one of the main findings of the study was that the primary phase had the highest impact within the mozzarella cheese supply chain. For the phases along the dairy chain, the mozzarella cheese making had the highest eco- efficiency ratio for each dairy chain and produced the highest economic value per unit of environmental impact. The milk production phase added the lowest value of total GVA in both dairy chains while showing the highest environmental impact in GHG terms. To reduce the environmental impact of the dairy chain and the wastage of a mozzarella cheese co-product, we assessed the carbon footprint of two dairy chains changing the diet composition at case farm level and using the liquid whey in cows' diet. The study hypothesis was that the use of the by-product of mozzarella cheese production within the local dairy chain would provide benefits under both environmental and economic perspectives. From the environmental point of view, the B supply chain with the whey showed an environmental performance per unit of mozzarella cheese lower than that of the A chain, although in a negligible measure, due to the effect of the milk yield in the primary phase. However, when considering the economic assessment of the two diets, the comparison of the eco-efficiency indicator evidenced a better performance of the B chain whose value per unit of impact was higher thanks to the liquid whey recycling. Ital. J. Food Sci., vol. 30, 2018 - 377 Study findings lead to certain conclusions on the need of improving both sides of sustainability. On the economic side, improvements are needed in the market mechanisms to set costs and revenues that increase the value added along the dairy chain, mainly at the farm level. Under an environmental perspective, based on the carbon footprint assessment, improvements in the milk production should provide practices and alternatives that can further reduce the primary phase emissions up to the limit allowed by the ruminant physiology. Finally, the circularity in nature and economic cycles should be further analysed to improve the performances of both sides of sustainability. By recycling the liquid whey and strengthening the relation between dairy farms and cheese factories at a local level, some economic benefits (the cost of whey transportation and the disposal costs of liquid whey) emerged, while the environmental burden of whey treatment is avoided. The best scenario satisfying both environmental and economic goals would realise a reduction in costs related to efficiency improvements in the usage of natural resources and dairy chain by-products, and a lower environmental burden associated with production processes. Concerning the revenues, the best scenario would be related to the attainment of a price premium for the environmental performances of the dairy products. For example by leveraging on marketing tools, such as environmental standards, labels, and environmental product declarations. ACKNOWLEDGEMENTS Authors are grateful for data generated by the INLATTE Project “Innovare naturalmente. Trasferimento di innovazione nella filiera lattiero-casearia per la valorizzazione del caciocavallo molisano e il recupero di sottoprodotti di lavorazione” University of Molise-Barone-Discenza-Molise Region. Regional Development Plan PSR 2007-2013, measure 124. REFERENCES Baitz M. 2017. Attributional Life Cycle Assessment. Ch. 3. In: “Goal and scope definition in life cycle assessment”. M.A. Curran (Ed.). LCA Compendium. The Complete World of Life Cycle Assessment. Springer Publishing, Dordrecht. Basset-Mens C., Ledgard S. and Boyes M. 2009. Eco-efficiency of intensification scenarios for milk production in New Zealand. Ecol. Econ. 68:1615-1625. Battini F., Agostini A. Tabaglio V. and Amaducci S. 2016. Environmental impacts of different dairy farming systems in the Po Valley. J. Clean. Prod. 112:91-102. Brereton M. 2006. Methodology Notes: Links between Gross Domestic Product (GDP) and Gross Value Added (GVA). Economic Trends 627:25-26. Brankatschk G. and Finkbeiner M. 2014. Application of the cereal unit in a new allocation procedure for agricultural life cycle assessments. J. Clean. Prod. 73:72-79. Colombini S., Zucali M., Rapetti L., Crovetto G.M., Sandrucci A. and Bava L. 2015. Substitution of corn silage with sorghum silages in lactating cow diets: in vivo methane emission and global warming potential of milk production. Agr. Systems 136:106-113. Dalla Riva A., Burek J., Kim D., Thoma G., Cassandro M. and De Marchi M. 2017. Environmental life cycle assessment of Italian mozzarella cheese: Hotspots and improvement opportunities. J. Dairy Sci. 100:1-20. Dalla Riva A., Kristensen T., De Marchi M., Kargo M., Jensen J. and Cassandro M. 2014. Carbon footprint from dairy farming system: comparison between Holstein and Jersey cattle in Italian circumstances. Acta Agr. Kapos. 18:75-80. EMEP/EEA. 2009. Air Pollutant Emission Inventory Guidebook. EEA. Technical report No 9/2009. http://www.eea.europa.eu//publications/emep-eea-emission-inventory-guidebook-2009 (accessed 21.07.17). Falcone G., De Luca A.I., Stillitano T., Iofrida N., Strano A., Piscopo A., Branca M.L. and Gulisano G. 2017. Shelf Life Extension to Reduce Food Losses: the Case of Mozzarella Cheese. Chem. Eng. Trans. 57:1849-1854. Ital. J. Food Sci., vol. 30, 2018 - 378 Falconi F., Neri P. and Olivieri G. 2011. Unpublished. Il ciclo di vita del latte con produzione di energia da biogas ottenuto da liquame, Doc. ENEA - PROT- RT–54. Fantin V., Buttol P., Pergreffi R. and Masoni P. 2012. Life cycle assessment of Italian high quality milk production. A comparison with an EPD study. J. Clean. Prod. 28:150-159. Fantozzi P., Caboni M.F., Gallina T.T., Gerbi V., Hidalgo A., Lavelli V., Perretti G., Pittia P., Pompei C., Rantsiou K., Rolle L., Sinigaglia M. and Zanoni B. 2015. Italy on the spotlight: EXPO Milan 2015 and Italian Journal of Food Science. It. J. Food Sc. 27(4):407-408. Finnegan W., Goggins J., Clifford E. and Zhan X. 2015. Global warming potential associated with dairy products in the Republic of Ireland. J. Clean. Prod. 1-12. Flysjo A., Thrane M. and Hermansen J.E. 2014. Method to assess the carbon footprint at product level in the dairy industry. Int. Dairy J. 34:86-92. Franchini F. and Neri P. 2004. Analisi del Ciclo di Vita di 1 L di latte UHT della ditta Granarolo, Doc. ENEA –PROT–INN 135–046. http://openarchive.enea.it//handle/10840/3825 (accessed 21.07.17) González-García S., Castanheira E.G., Dias A.C. and Arroja L. 2013. Environmental performance of a Portuguese mature cheese-making dairy mill. J. Clean. Prod. 41:65-73. Hawkins J., Weersink A., Wagner-Riddle C. and Fox G. 2015. Optimizing ration formulation as a strategy for greenhouse gas mitigation in intensive dairy production systems. Agr. Syst. 137:1-11. Helmes R., Ponsioen T. and Robbemond R.M. 2016. Allocation choices strongly affect technology evaluation in dairy processing. Paper presented at the “Putting LCA into practice”. 10th International Conference on Life Cycle Assessment of Food”, Dublin, October 19- 21. Hessle A., Bertilsson J., Stenberg B., Kumm K-I. and Sonesson U. 2017. Combining environmentally and economically sustainable dairy and beef production in Sweden. Agr. Syst. 156:105-114. Huysveld S., Van Meensel J., Van linden V., De Meester S., Peirena N., Muylle H., Dewulf J. and Lauwers L. 2017. Communicative farm-specific diagnosis of potential simultaneous savings in costs and natural resource demand of feed on dairy farms. Agr. Syst. 150:34-45. IDF. 2015. A common carbon footprint approach for the dairy sector - The IDF guide to standard life cycle assessment methodology. No. 479/2015. http://www.fil-idf.org/wp-content/uploads/2016/09/Bulletin479-2015_A-common- carbon-footprint-approach-for-the-dairy-sector.CAT.pdf. (accessed 21.07.17). INLATTE Project. OR.2. “Validazione dell’impiego di siero, stabilizzato per via fermentativa, nell’alimentazione di bovine da latte”. Salimei E., Zurlo, N. http://www.premioimpresambiente.it/wp- content/uploads/allegati/390/Allegato-2-Salimei-Zurlo.pdf. IPCC. 2006. Guidelines for National Greenhouse Gas Inventory. Volume 4. Agriculture, Forestry and Other Land Use. Chapters 10-11. http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (accessed 12.05.17). IPCC. 2015. Climate Change 2014. Intergovernmental Panel on Climate Change, Synthesis Report. https://www.ipcc.ch/pdf/assessment-report/ar5/syr/SYR_AR5_FINAL_full_wcover.pdf (accessed 12.05.17). Iribarren D., Hospido A., Moreira M.T. and Feijoo G. 2011. Benchmarking environmental and operational parameters through eco-efficiency criteria for dairy farms. Sci. Total Environ. 409:1786-1798. ISPRA. 2008. Agricoltura - Inventario nazionale delle emissioni e disaggregazione provinciale. Rapporto 85/2008. http://www.isprambiente.gov.it/contentfiles/00003600/3620-rapporto-85-2008-inventario-nazionale-agricoltura- alta.pdf/view (accessed 12.05.17). Kim D., Thoma G., Nutter D., Milani F., Ulrich R. and Norris G. 2013. Life cycle assessment of cheese and whey production in the USA. Int. J. Life Cycle Assess. 18:1019-1035. Kirilova E.G. and Vaklieva-Bancheva N.G. 2017. Environmentally friendly management of dairy supply chain for designing a green products' portfolio. J. Clean. Prod. 167:493-504. Koesling M., Hansenc S. and Azzaroli Bleken M. 2017. Variations in nitrogen utilisation on conventional and organic dairy farms in Norway. Agr. Syst. 157:11-21. Ital. J. Food Sci., vol. 30, 2018 - 379 Meul M., Nevens F., Verbruggen I., Reheul D. and Hofman G. 2007. Operationalising eco-efficiency in agriculture: the example of specialized dairy farms in Flanders. Prog. Ind. Ecology. Int. J. 4:1-2. Mirabella N., Castellani V. and Sala S. 2014. Current options for the valorization of food manufacturing waste: a review. J. Clean. Prod. 65:28-41. Mu W., van Middelaar C.E., Bloemhof J.M., Engel B. and de Boer I.J.M. 2017. Benchmarking the environmental performance of specialized milk production systems: selection of a set of indicators. Ecol. Ind. 72: 91-98. Muller K., Holmes A., Deurer M. and Clothier B.E. 2015. Eco-efficiency as a sustainability measure for kiwifruit production in New Zealand. J. Clean. Prod. 106:333-342. Murphy B., Crosson P., Kelly A.K. R. and Prendiville R. 2017. An economic and greenhouse gas emissions evaluation of pasture-based dairy calf-to-beef production systems. Agr. Syst. 154: 124-132. Nemecek T., Schmid A., Alig M., Schnebli K. and Vaihinger M. 2011. Variability of the Global Warming Potential and Energy Demand of Swiss Cheese. In Proceedings of SETAC Europe 17th LCA Case Studies Symposium “Sustainable Lifestyles”, Budapest, Hungary, 28 February - 1 March. Neri P. (Ed.) 2009. L’analisi ambientale dei prodotti agroalimentari con il metodo del life cycle assessment. ARPA Sicilia - Agenzia Regionale per la Protezione dell’Ambiente della Sicilia. ARPA Strumenti. http://www.arpa.sicilia.it/attivita/pubblicazioni/ (accessed 12.05.17). Neri P. and Borsari A. 2005. Analisi ambientale del ciclo di vita della produzione di latte da allevamento biologico e confronto con la convenzionale, Doc. ENEA –PROT–135–086. http://openarchive.enea.it/handle/10840/3859 (accessed 12.05.17). Notarnicola B., Salomone R., Petti L., Renzulli P.A., Roma R. and Cerutti A.K. (Eds.). 2015. “Life cycle assessment in the agri-food sector. Case studies, methodological issues and best practices”. Springer International Publishing, Switzerland. Pagani M., Vittuari M., Johnson T.G. and De Menna F. 2016. An assessment of the energy footprint of dairy farms in Missouri and Emilia-Romagna. Agr. Syst. 145:116-126. Palmieri N., Forleo M.B. and Salimei E. 2017. Environmental impacts of a dairy cheese chain including whey feeding: an Italian case study. J. Clean. Prod. 140:881-889. Patra A.K. 2012. Enteric methane mitigation technologies for ruminant livestock: a synthesis of current research and future directions. Environ. Monit. Assess. 184:1929-1952. Prazeres A.R., Carvalho F. and Rivas J. 2012. Cheese whey management: a review. J. Environ. Manage. 110:48-68. Roma R., Corrado S., De Boni A., Forleo M.B., Fantin V., Moretti M., Palmieri N., Vitali A. and De Camillis C. 2015. Life cycle assessment in the livestock and derived edible products sector. Ch. 5. In: “Life cycle assessment in the agri-food sector: case studies, methodological issues and best practices”. B. Notarnicola, R. Salomone, L. Pett, P.A. Renzulli R. Roma A.K. Cerutti (Eds.). Springer Publishing, Switzerland. Röös E., Sundeberg C. and Hansson P.A. 2014. Carbon footprint of food products. Volume 1. Ch. 4. In “Assessment of carbon footprint in different industrial sectors”. S. S. Muthu (Ed.). Springer Publishing, Singapore. Rotz C.A. 2018. Modeling greenhouse gas emissions from dairy farms. J. of Dairy Sci. 101:1-16. Sala S., Anton A., McLaren S.J., Notarnicola B., Saouter E. and Sonesson U. 2017. In quest of reducing the environmental impacts of food production and consumption. J. Clean. Prod. 140:2387-2398. Saling P. 2016. Eco-efficiency assessment. Ch. 4. In “Special types of life cycle assessment part of the series LCA compendium – the complete world of life cycle assessment”. M. Finkbeiner (Ed.). Springer Publishing, Dordrecht. Sanjuan N., Ribal J., Clemente G. and Fenollosa M.L. 2011. Measuring and improving eco-efficiency using data envelopment analysis a case study of Mahon-Menorca cheese. J. Ind. Ecol. 15:614-628. Santos H.C.M., Maranduba H.L., de Almeida Neto J.A. and Rodrigues L.B. 2016. Life cycle assessment of cheese production process in a small-sized dairy industry in Brazil. Environ. Sci. Pollut. Res. 1:1-13. Soteriades A.D., Faverdin P., Moreau S., Charroin T., Blanchard M. and Stott A.W. 2016. An approach to holistically assess (dairy) farm eco-efficiency by combining Life Cycle Analysis with Data Envelopment Analysis models and methodologies, Animal 1-12. Succi G., Crovetto G.M. and Salimei E. 1986. Alimentazione liquida nella vacca da latte. Agricoltura 2000. 4: 12-17. van Middelaar C.E., Berentsen P.B.M., Dijkstra J., and de Boer I.J.M. 2013. Evaluation of a feeding strategy to reduce greenhouse gas emissions from dairy farming: the level of analysis matters. Agr. Syst. 121:9-22. Ital. J. Food Sci., vol. 30, 2018 - 380 van Middelaar C.E., Berentsen P.B.M., Dolman M.A. and de Boer I.J.M. 2011. Eco-efficiency in the production chain of Dutch semi-hard cheese. Livest. Sci. 139:91-99. Vergé X.P.C., Maxime D., Dyer J.A., Desjardins R.L., Arcand Y. and Vanderzaag A. 2013. Carbon footprint of Canadian dairy products: calculations and issues. J. Dairy Sci. 96:6091-6104. Vida E. and Tedesco D.E.A. 2017. The carbon footprint of integrated milk production and renewable energy systems. A case study. Sci. Total Envir. 609:1286-1294. WBCSD. 2000. Measuring eco-efficiency: a guide to reporting company performance. https://www.gdrc.org/sustbiz/wbcsd.html (accessed 12.05.17). Weidema B.P., Bauer C., Hischier R., Mutel C., Nemecek T., Reinhard J., Vadenbo C.O. and Wernet G. 2013. Overview and methodology: Data quality guideline for the ecoinvent database version 3. Swiss Centre for Life Cycle Inventories. Ecoinvent Report N. 1, Vol. 3. http://vbn.aau.dk/ws/files/176769045/Overview_and_methodology.pdf (accessed 18.05.17). Wettemann P.J.C. and Latacz-Lohmann U. 2017. An efficiency-based concept to assess potential cost and greenhouse gas savings on German dairy farms. Agr. Syst. 152:27-37. White R.R. 2016. Increasing energy and protein use efficiency improves opportunities to decrease land use, water use, and greenhouse gas emissions from dairy production. Agr. Syst. 146: 0-29. Paper Received November 13, 2017 Accepted December 30, 2017