Allocation factors for meat coproducts: Dataset to perform life cycle assessment at slaughterhouse

The sharing of total environmental impacts between the different products of a multi-output system is crucial in Life Cycle Assessment. ISO standards recommend subdivision then substitution methods when possible. Sometimes, allocations rules are necessary. They consist of allocating the total impact to the different products in proportion to a value that characterize the products. They can be based on physical parameters (such as mass, protein, dry matter, etc.) or the economic value of coproducts can be used as a proxy. As they are based on various type of parameters, allocation rules can lead to significantly different environmental impact results. Then a consensus is difficult to reach between stakeholders as for example in meat sector. To make the debate going further, Chen et al. (2017) proposed a new allocation method based on biophysical parameters (Chen et al., 2017). Adapted from previous methods, they propose to allocate impacts in proportion to the energy needed for the growth, the maintenance and the activity of each tissue. The method has been judged as scientifically viable but also particularly difficult to apply due to the amount of necessary data and to the complexity of the calculation model. In a recent project, we developed a freeware to easily calculate biophysical allocation factors as well as mass and economic factors to allow a fair comparison: MeatPartTool. We also collected data to create a dataset of mass, economic and biophysical allocation factors for a large range of beef (132 individuals), calf (54 individuals) and lamb (14 individuals) at the slaughterhouse stage. This data paper provides both primary data and calculated allocation factors.


a b s t r a c t
The sharing of total environmental impacts between the different products of a multi-output system is crucial in Life Cycle Assessment. ISO standards recommend subdivision then substitution methods when possible. Sometimes, allocations rules are necessary. They consist of allocating the total impact to the different products in proportion to a value that characterize the products. They can be based on physical parameters (such as mass, protein, dry matter, etc.) or the economic value of coproducts can be used as a proxy. As they are based on various type of parameters, allocation rules can lead to significantly different environmental impact results. Then a consensus is difficult to reach between stakeholders as for example in meat sector. To make the debate going further, Chen et al. (2017) proposed a new allocation method based on biophysical parameters (Chen et al., 2017). Adapted from previous methods, they propose to allocate impacts in proportion to the energy needed for the growth, the maintenance and the activity of each tissue. The method has been judged as scientifically viable but also particularly difficult to apply due to the amount of necessary data and to the complexity of the calculation model. In a recent project, we de-veloped a freeware to easily calculate biophysical allocation factors as well as mass and economic factors to allow a fair comparison: MeatPartTool. We also collected data to create a dataset of mass, economic and biophysical allocation factors for a large range of beef (132 individuals), calf (54 individuals) and lamb (14 individuals) at the slaughterhouse stage. This data paper provides both primary data and calculated allocation factors.
© 2020 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ) Specifications Table   Subject Environmental Science -Environmental Impact Assessment Specific subject area Allocation factors for Life Cycle Assessment of meat coproducts Type of data

Value of the Data
• These data are useful as they move the debate on allocation factors for LCA of meat forward. There is no consensus between stakeholders on the subject when choosing between allocation methods. The lack of data however makes methods difficult to use and compare. Here is proposed an unprecedented range of mass, economic and biophysical allocation factors for meat coproducts. • This dataset will benefit to everyone who wants to practice LCA to meat products at slaughter stage. Researchers, industrials, decision-makers are interested to better understand environmental impacts of meat. If they cannot calculate their own allocation factors, they can pick the most appropriate ones in this dataset. Furthermore, by proposing both mass, economic and biophysical allocation factors, the dataset sets political questions aside but offers material to discuss.
• These data can be directly used to allocate environmental impacts between meat coproducts at slaughter stage. The vast range of individuals proposed allows the user to choose appropriate allocation factors instead of generic ones. Furthermore, as both mass, economic and biophysical allocation factors are calculated, the user will be able to easily provide sensitivity analysis when using one or another. • This dataset offers a large range of mass, economic and biophysical allocation factors that were not available so far in literature. From now, only a few ones existed, mostly generic cases. This is the beginning towards more differentiated datasets appropriated to different realities. The authors think that this dataset should be completed by other individuals, especially from different geographical areas. To help, we developed a freeware that calculates mass, economic and biophysical allocation factors by mixing input data provided by the user and possibly default data if the user miss some.

Input data
Primary data (i.e. all the dataset necessary to calculate allocation factors) have been collected from different sources: literature, previous projects and expert interviews. In total, the dataset comprises 132 beef, 54 calves and 14 lambs ( Table 1 ).
For each species, the list of coproducts has been drawn up. For a given species, it is considered that every breed comprises the same coproducts. Each coproduct is then classified by: -C1 products are those that presents risks of: • Spongiform encephalopathy transmission; • Presence of residues of toxic substances; • Presence of environmental contaminants. -C2 products are those coming from digestive system that present health hazards -C3 products are free of risks and used as intrants for industrial production (for example petfood or fertilizers) C1 and C2 products are generally discarded.
Lists of coproducts and associated destinations and groups of tissues for bovine, calf and ovine are respectively available in Tables 2-4 . These tables also contain, for each coproduct, the percentages of Water, Dry Matter, Lipids and Proteins. These are the same for every breed of a given species in the present dataset. Those data concerning quantity of coproducts and their physicochemical compositions were compiled by Gac et al. (2012) considering bibliographic references, supplemented by extrapolations and expert estimates when information was lacking.
For each coproduct, the mass fraction of the total mass is necessary. It has been calculated for each breed of each species. They can be considered as generic data to characterize coproducts. Data from Gac et al. (2012) are used as a reference and adapted to each breed depending on carcass yields [2] .

mass f raction of the coproduct i f rom breed j
Carcass Yields are available in Tables 5-7 . They come from Laisse et al. [3] . These table also contains the Empty Body Weight at slaughter age that differs from a breed to another. These data have been obtained on the basis of a census data extraction operated by Institut de l'Elevage (GES Division) from SPIE (the Professional Livestock Information System approved by the French State), which contains data from the BDNI (National Data Base of Identification which register all animal birth and movements), completed by the Normabev database (concerning slaughtering of bovines). This French information system on livestock is described by Delomel and Gibon [4] . When data were not available, mean values have been used.
Next table contains a list of parameters that are identical for each breed of a given species ( Table 8 ). These parameters are used by the model developed by Chen et al. [1] to calculate the Table 2 Destinations, group of tissues and composition of beef coproducts (from Gac et al. (2012) Others  72  28  10  16  Tonsil  C1-C2 for disposal  Others  75  25  12  10  Trachea  Pet food  Others  65  35  5  29  Udder  Pet food  Others  86  14  5  3  Upper throat  Pet food  Others  70  30  5 20 Water in the rumen Spreading/Compost GIT 99 1 0 3 Finally, a coefficient is used to modulate the energy required for the activity. These coefficients are specific for breeds and depend on the rearing mode. Data from IPCC (2006) are used [6] . Data are available in Table 9 . To calculate economic allocation factors, an economic dataset has been built by compiling data from ACYVIA [7] . The dataset is available respectively for beef, calf and lamb in Tables 10-12 All these input data are also available in a complete * .csv file (supplementary file 13). This is the formatted database as used by MeatPartTool calculation freeware.

Allocation factors
For each individual, mass (based on wet mass), economic and biophysical allocation factors are given per kg of coproduct. They are respectively available for bovine, calf and ovine in supplementary file 1, supplementary file 4 and supplementary file 7. Then the total weightings by coproduct (i.e. allocation factor per kg multiplied by the mass of coproduct) are also given (respectively available in supplementary files 2, 5 and 8). Finally, an aggregation by destination category (e.g. Human food, PAP C3, etc.) is also available (respectively in supplementary files 3, 6 and 9).

Experimental Design, Materials and Methods
Mass and economic allocation factors have been calculated by following LCA standards. Biophysical allocation factors calculation was performed using Chen et al. (2017) model. A calculation freeware has been developed in Python. The code section that concerns the calculation are given in supplementary files 10, 11 and 12. A specific code is used for each species. These are Python files readable with any code editor (as Notepad ++ ). They work with extra code, formatting a list from a * csv.file. The complete code is implemented in the MeatPartTool open-source freeware [8] .
One at a Time sensitivity analysis is provided for the two variant input parameters. The variation of the share of human food destination coproducts is given when testing different Gompertz Coefficients, Carcass Yields and Rearing methods. Results are summed up in Table 13 and more details are provided in Supplementary Files 14. Results are the most sensitive to Gompertz Coefficient with only 10% of variation between extreme values. Very few information was found about this parameter in the case of the present study. Consequently, the authors think that biophysical allocation would benefit from more research on Gompertz coefficient in the future.

Declaration of Competing Interest
This work received financial support from Interbev (SECU 19-20).