The IPCC [2] recommended GHG factors for CO₂ (1), CH₄ (23) and N₂O (296) to calculate CF for various processes. Recently there was some discussion about the factors and the IPCC [49] revised the global warming potential of methane (CH₄) from 23 to 25 because of indirect effects of CH₄ on ozone and stratospheric water vapour [9]. On the other hand, the Worldwatch Institute [50] suggests a Global Warming Potential for CH₄ of 72. Methane is a very important gas for CF calculation, especially for food derived from ruminants. Between 50 and 80% of the total GHG emissions of food of ruminant origin are due to methane [37,38].

The N₂O-factor is given with around 300 (296 [2]; 298 [49]). From a global view the agricultural N₂O-emissions (from manure and soil; given in billion tonnes of CO_2eq/y) are larger than calculated methane emissions (2.5 versus 2.15 billion t CO_2eq/y, [6]). Analogue tendencies are reported from agriculture in the USA (222 million t CO_2eq/y from N₂O; 197 million t CO_2eq/y from CH₄; [51]). From the view of science and policy, further research is required to considering a time horizon of the GHG emissions [9,52].

In addition to GHG factors a possible exact measuring of climate relevant gases in all links along the food chain is an essential prerequisite to calculate CF.

There is general agreement that carbon dioxide emissions from livestock metabolism are not considered as a CO₂-source in CF ([2,9], see Figure 1). CO₂ has been fixed by photosynthesis in phytogenic biomass and the C consumed in feed and emitted as CO₂ by animals is considered as equivalent or as emission neutral. On the other hand, CO₂ from technical processes associated with animal husbandry should be considered in CF calculations (for details see [7,10,11]).

Methane can be considered as an unavoidable by-product of anaerobic microbial fermentation, especially in the rumen of ruminants, but also in the hindgut of all species and during anaerobic manure management. Since energy losses via methane in the rumen are well known [53], animal nutritionists have been trying to reduce the gastrointestinal methane emission from ruminants and in the hindgut of further species for a long time. In 2005 around 90 million tonnes CH₄ (about 1.9 billion t CO₂eq/y) were emitted from gastrointestinal fermentation of ruminants [6]. These emissions are projected to grow by over 30% from 2000 to 2020 [6]. Enteric methane emissions (e.g., [54,55,56,57]), methods of measurements (e.g., [58,59,60,61]) and reduction potentials (e.g., [62,63,64,65]) should not be further considered here. Furthermore there are special issues of the journals “Animal Feed Science and Technology” 145 (2008), pp. 209–419 and 166–167 (2011), pp. 1–782 and in the “Australian Journal of Experimental Agriculture” 48, No. 1 and 2 (2008), dealing with the topics mentioned above.

Laughing gas (N₂0) has the highest GHG potential from the most relevant GHG [2]. It is not directly excreted by animals; it depends on microbial conditions during manure management and in the soil. Lower N-excretion by animals [66,67,68] and improved N-management (e.g., storing of manure, adequate amounts of manure and fertilizer) may contribute to lower N-emissions from manure and soil (e.g., [69,70,71,72]). More representative results from N₂O measurements (also for grazing animals) may substantially contribute to more reliable data for calculation of CF along the food chain. In some cases specific emissions were excluded from calculations of CF due to data uncertainty (e.g., N₂O from leguminous pastures [37]). Such special situations should be mentioned in the study report to assess the results.

Definition of system boundaries along the food chain (see Figure 1 and [24]) is the starting point for GHG measurements, for the calculation of CF of livestock products and for comparing results from various studies [4,10]. There are some open questions which need to be answered and it should be clearly defined whether they are or are not considered in the calculation, such as:

For example under Australian conditions Peters et al. [37] defined the system boundary in the case of beef for all on-site and upstream processes at the farm, feedlot, and whole processing plant, including transport between these sites.

Table 3 shows exemplary CF for milk under consideration of various boundaries. A clear definition of the system boundaries and the comprehensibility are important prerequisites to follow the calculations and to make the results comparable (e.g., [37,41,44]). Scientists working in this field should come to agreements concerning system boundaries and GHG factors of climate relevant gases.

Milk and eggs are clearly defined as food of animal origin. The yield can be measured as weight (kg, etc.) or on the basis of standardized products (e.g., standardized protein, fat, dry matter or energy). Therefore it is relatively easy to measure the animal yield for further calculations. But nevertheless, there is a certain range between CF for milk (see Table 1).

The composition of milk and eggs is well defined (see Table 6), but it may vary between various sources and depending on animal breed, feeding and other influencing factors. Therefore the analysis of milk and egg composition (protein; fat, lactose) may contribute to being more specific in measuring the animal yield incl. the energy yield. Milk and eggs may be used entirely as food (except small amounts of colostrum and egg shells; see Table 7).

It is much more difficult to quantify and characterize the yield from the animal body after slaughtering and processing. The GHG balance per kilogram body weight gain can only be calculated on the farm level. Mialon et al. [82] carried out a feeding study with Blond d’Àquitaine bulls in the finishing phase (400–650 kg body weight) including various feeding systems and weight gains between 1,494 and 1,862 g/d and Doreau et al. [83] calculated a CF between 3.6 and 4.7 kg CO_2eq per kg body weight gain. Those values are similar to high daily weight gains as shown in Table 5, but much lower than for lower weight gains. Normally, the GHG emissions for the whole beef system include also emissions of cows, calves and heifers, needed to produce beef. They are much higher than in the system dairy cow—growing/fattening bulls for beef (see allocation).

Mostly the term “meat” is used, but it is not clearly described, what it means (real meat or meat plus bones). Peters et al. [37] introduced the term “hot standard carcass weight” (HSCW) as the weight at the exit gate of the meat processing plant. It varies between 50–62% of the live weight of the cattle to be slaughtered, but it may vary between 50% in the case of sheep and up to 80% for fattening turkeys (e.g., [12,37,83]).

In the case of animals for meat/fish production the following endpoints can be measured:

Therefore it is really difficult to find an adequate CF for meat or edible products from slaughtered animals. Various authors used different bases to calculate CF for products from slaughtered animals. Williams et al. [12] estimated the killing out percentages for beef and poultry with 55 and 70% and 72, 75 and 77% for pigs with live weights of 76, 87 and 109 kg, respectively. Lesschen et al. [7] used fixed values to calculate the carcass fraction from the final body weight of animals (e.g., 58% for beef; 75% for pork and 71% for poultry). Most authors used a fixed fraction of 0.9 for all animal species for conversion of carcass weight to edible “meat”. De Vries and de Boer [5] used calculation factors to determine the amount of edible product per kg live weight by 0.43; 0.53 and 0.56 for beef, pork and poultry. Table 4 shows potential outputs for growing/fattening cattle under consideration of various endpoints as mentioned above.

Calculation of CF may base on various outputs. For practical reasons carcass weight or weight gain (warm or cold) would be the most important endpoint to measure the yield of slaughtered animals because this weight is measurable in the abattoir [37] and can be used for further calculations. Based on the values derived from Table 4, CF is calculated for various endpoints under consideration of differences in feeding and greenhouse gas emissions and is shown in Table 5.

The production of protein of animal origin is one of the most important goals of animal husbandry [5]. On the other hand, the efficiency and the emissions of animal products can be also compared on the basis of edible protein. The N or protein content of various foods of animal origin may vary from values used for calculations in Table 6 (data by [84] on the basis of own studies). Our data agrees with values used by Lesschen et al. [7], and it does not substantially disagree with values from human food tables (see Table 6). De Vries and de Boer [5] used for their calculations 190 g protein/kg edible beef, pork and poultry meat; 30 g per kg milk products and 130 g per kg eggs.

Considering various influencing factors such as animal yields, feeding, edible fractions and protein content in the edible fractions, the yield of edible protein per day and per kg body weight of animals is given in Table 7.

The feeding may influence CF of food of animal origin. In the case of ruminants, higher amounts of concentrate are required for higher animal yields. The proportion of by-products [90,91] used in animal feeding does not only have nutritional implications, but it also affects the results of calculations on land use [92]. There are large differences in protein yield per animal per day or per kg body weight and day depending on animal species and category as well as their performances and the fractions considered as edible (see Table 7).

Table 7 shows the highest protein yields per kg body weight for growing broilers as well as for laying and lactating animals and the lowest values for growing/fattening ruminants. Based on those values, emissions per kg edible protein are given in Table 8. Higher proportions of edible fractions or higher protein content (e.g., 50 g protein per kg camel milk, [77]) as shown in Table 6 and Table 7 may increase the protein yield and reduce the CF per unit of product.

Apart from protein food of animal origin also contains fat and some carbohydrates which contribute to human nutrition and which may replace energy of plant origin in human diets.

At high levels of performance there are remarkable differences in CO₂ emissions due to human consumption of 1 g protein from food of animal origin (eggs and meat from broiler < pork < milk < beef, see Table 7). But here it has to be emphasized that this protein intake is accompanied—willingly or not—by an energy intake from the protein itself, but also from further nutrients like lactose and fat in milk or from fat in meat or eggs. Therefore, it should be avoided to attribute the CO₂ burden to the protein fraction (“edible protein”) exclusively. To prevent that this fact is neglected, there are different alternatives:

In a first simple method, the CO₂ emission due to 1 kg edible protein could be used as CO₂ burden of consumed energy (for example: 1 kg edible protein of eggs corresponds to about 8 kg egg corresponding to 51.6 MJ energy (calculated by [77]); these combined intakes are related to 3 kg CO₂).

One alternative could be a “nutritional allocation” (as described before for economic allocation), meaning that the CO₂ emissions are attributed to different functions of the food (source of protein/source of energy/source of further essential nutrients).

In a first simple step it is recommended to diminish the CO₂ emission per 1 kg edible protein (Table 8) by the CO₂ amounts that would occur at an identical energy intake from food of plants (energy from carbohydrates and fat). It means that an intake of 1 kg protein from eggs (corresponds to 8 kg eggs; see Table 7; and corresponding to 51.6 MJ energy) saves high amounts of other food (and their CO₂ burden). A more sophisticated way of an “allocation” within the foods could be to differentiate between “protein derived energy” and “non-protein derived energy”. In milk and eggs more than 50% of the total energy content is related to the non-protein-fraction (lactose/fat), therefore, it is questionable whether the entire CO₂ emission should be attributed only to the protein intake. Due to the very low CO₂ emission caused by energy intake of carbohydrates and fat from plants/seeds [12,15,17] it would avoid/save high amounts of CO₂ emissions, if the production of food of animal origin focussed on “edible protein” and not on energy of non-proteinaceous fractions.

Furthermore animal products are not only used as food or respectively, as protein/amino acids, and energy sources; they also offer some other important side-products such as skins or hides, fish meal or meat and bone meal, etc. A kind of combined “nutritional/further purposes allocation” may contribute to a more scientific assessment of CF for nutrient and energy supply as well as further uses.

Advantages and weaknesses of endpoints (outputs) of various types of animal production are summarized in Table 9. All endpoints are characterized by some advantages and disadvantages. From nutritional and scientific points of view the edible protein seems to be the most favourable measurement, but its measurement is not easy and requires some analytical work (see Table 9). Land requirements (e.g., arable land and/or grassland) as well types and intensities of food production may be calculated on the basis of various protein sources for human nutrition. Such calculations can contribute for better understanding of various conflicting aims in the field of food production, human nutrition, use of unlimited and limited resources and resource efficiency, emissions and further points in public discussion.